Nishi Family › Compare › Workflow Automation

Nishi Compare · capabilities, evidence and delivery

Workflow Automation

Compare capabilities and delivery evidence. Source discovery, qualification, deployment and accepted outcomes are separate; an open contract names work still to verify.

Nishi vs Zapier and n8n and Temporal and Microsoft Power Automate

Overview

The purpose, declared position and evidence coverage of this domain. Source presence and completed acceptance are different measures.

Where we are. The sovereign workflow engine is real and deep: event-sourced run ledger with per-step retry and crash-resume (R1), exec-organ connector catalog (R3), human approval steps (R4), run observability by ledger replay (R6), a certified console (R7), variables and substitution incl forked argv (R8), conditions (R9), on-failure routing (R10), for-each (R11), versioning with drift witness (R12) and a template gallery (R13); plus the rules engine, tickless RRULE scheduler, reminders, multi-channel send incl physical mail, docgen, never-brick ops pipeline and SSE event stream. Two structural exceeds: zero per-task cost on own stack and tamper-evident never-lose run history. Open: every INBOUND and OUTBOUND door the incumbents are defined by (webhooks, SaaS connectors), the visual canvas and its interaction UX, the NL copilot, RPA and the macro recorder.

Where we need to go. An automation product a non-programmer can drive end to end on owned hardware: a webhook in, a SaaS call out, a canvas to draw it, a copilot to draft it, a recorder to capture it -- with every run still an additive ledger and every step still an organ. Temporal-class durability is already ours; Zapier-class reach is the climb.

The unit. 1 u = one measured session-leg, the estate's shared calibration (graphics R21 in one leg 2026-08-15). Local calibration: R8 through R13 each landed gated GREEN as one leg apiece in the engine lane -- estimates below are relative to those.

Where we are: 13 landed, 8 open. Engine depth at parity with Temporal on the ledger axes; reach and product UX behind Zapier and n8n. Counts measured at emit below this line.

Cost to the doors: 2 u. WA1 inbound webhooks and WA2 the outbound connector call -- the two rungs that turn an internal engine into an integration product.

Cost to a visual product: 5 u. WA3 the canvas emitter, WA4 inline test and edit-in-place, WA5 onboarding and docs -- the Zapier editor bar, measured by the existing ui-judge.

Cost to exceed on reach: 9 u. WA6 the NL-to-flow copilot over the sovereign model, WA7 RPA UI steps and WA8 the record-and-replay recorder -- Power Automate Desktop's category, on our own stack.

Research bar. Zapier is measured on webhooks, 8000-plus app connectors, Copilot builder, editor UX. Theirs: the category definition for reach and editor. Ours: WA1 WA2 WA4 WA6 measured on this page.

Research bar. n8n is measured on open-source node canvas, IF/Switch, loops, AI nodes. Theirs: the open-source canvas reference. Ours: WA3 and WA4.

Research bar. Temporal is measured on durable execution, timers, signals, child workflows, visibility. Theirs: depth of the durable engine. Ours: already matched on the ledger axes; timers and signals remain depth we concede.

Research bar. Power Automate Desktop is measured on RPA, macro recorder, approvals UI. Theirs: the desktop-automation bar. Ours: WA7 and WA8.

Latest recorded release

2026-09-30 · R1

2026-09-30 ENABLEMENT: the existing Nishi shipping owner now authors and consumes native NGP1 per-gate inputs. Installed nx_organ_ship329176B SHA31f1ea03162326481b19e78db3bda563cb3cdba70941dbb65efeccbd03143295 has independently matched serving/runner copies. New gate62/62 and existing71/71 plus151/151 passed. Installed author replay5249 retained the same1368B native recipe172a5440 with no duplicate; real consumer build1790785287 used13 inspection inputs and8 definition inputs separately, passed54 plus41+4 checks, and promoted nx_pm_owned_define884592B SHAb44f48328efdc001bc7cbb993bec96083c0223bb5916cf49c60efebef3082b5b with independent serving/runner equality. Exact input/config identities, required-gate coverage, literal arguments and per-execution ownership are enforced; malformed, conflicting and over-budget records refuse. Source-closure banking6066 produced byte-identical qualified native gate79928f66; current closure independently checked6384. The definition gate now emits its canonical verdict after all cleanup, qualified47/47 in6334 and accepted by the original build API6383 as375600B SHAa803cfd5. Its100 isolated test files/21686B were removed and directory absence independently confirmed. These gate artifacts are qualification-only; no separate gate installation claimed. Discovery row refreshed4158; provider caches unverified. Detailed retained worklog buildroot/runtime/clmem/project-owned-alpha-native-architecture-2026-09-28.md names original evidence, rollback copies, superseded candidates and fixture debt. OPEN: automatic permanent gate workloads with legacy-call compatibility, old shipping-gate fixture disposition, full PM-owned repair/build/verify/recover/close, credential retirement, original incident closure and comparative resource return. Whole-conf recipe invalidation remains strict. Browser execution-host/config integration and served visual/interaction review remain OPEN. This is one installed and exercised build-composition capability in alpha, not autonomous recovery, estate-wide SOTA, Beta or measured net savings.

Release history and work log

Release entries describe recorded changes; they do not establish that every capability passed evaluation.

26 of 34 capability source symbols found|2 of them authored stronger-claims|8 open|coverage 764/1000|adoption 21 full / 5 partial

Evidence profile — what the gaps on this board actually are

Measured by nx_swcompare_evidence, read back by nx_evprofile_lib. Every figure is a count with its denominator — there is deliberately no score, no grade and no percentage anywhere in this band, because a stored scalar is a field a seat can edit and a counted partition is not.

proven able to fail counts the gates that have a RECORDED RED — nx_gate_bite mutated the gate subject, rebuilt it, watched the gate go red, and that record is inside the shared TTL. never bitten is its complement over the same denominator: those gates ran and were green, and nothing has ever shown them able to detect anything, so their green is a statement about this run and not about the gate. green at 0/0 is a separate and much weaker observation — the gate printed GREEN on a zero denominator, so its own tooth counter says it examined nothing. A gate can be green, non-zero, and still never bitten; that is the common case and it is now visible instead of implied.

partition: grounded + unsupported = 26 vs present 26 · named + unnamed + withheld = 8 vs open 8 · both reconcile

liar-kill conj=GPQN · all four conjuncts held

graded document: BUILDROOT tree, 7357 bytes · gates map: PRIMARY · stamped 1d 7h ago · source ../knowledge/status/evstamp_automation.verdict

Gap class	What it is, and the work it names
NONE	No gap class fires on this board: every published claim is grounded, the declared gates ran and were green, no watch contract is unnamed or already landed, and the stamp is inside its TTL. That is a statement about evidence, not about depth, scale or polish.

This band reports the referee counts. The per-axis worklist rows — which axis is unsupported, which watch is ready to flip, which gap is unnamed — are printed by nx_swcompare_evidence automation itself and are not carried on the stamp, so this page names the classes and the producer names the rows. That split is stated rather than hidden: a count without a worklist is not actionable, and this band is honest about which half of that it is.

Production map

Follow the dependencies, declared acceptance criteria and recorded priorities. Inspect source binding before treating a rank as executable work.

Ranking source binding: STALE_OR_INVALID_BINDING. Recorded priorities require current acceptance evidence and resource checks before execution.

Recorded ranking withheld

Retained ranking is not bound to the current plan and matrix: STALE_OR_INVALID_BINDING. Historical data remains retained. Refresh the ranking after resolving the input differences. Latest execution outcome is unverified.

Declared roadmap — contract, acceptance, executor, effort

Rung	Closes with	Definition of done (pre-declared)	Executor	Est.
Durable workflow engine (R1)	wf_resume	Event-sourced run ledger, per-step retry, durable dedup, crash-resume -- LANDED gated GREEN	Organ	0 u
Organ connector catalog (R3) after R1	wf_exec_organ	Exec-organ over a fail-closed catalog -- LANDED gated GREEN	Organ	0 u
Approval steps (R4) after R1	wf_decide	Park until an APPROVED or DENIED ledger decision lands, deny wins -- LANDED	Organ	0 u
Variables and substitution (R8) after R1	wf_subst	Event fields bind as run vars, connector stdout captured, placeholder substitution incl forked argv -- LANDED	Organ	0 u
Conditions (R9) after R8	wf_cond_ok	Per-step k=v condition vs the variable plane, mismatch SKIPs recorded -- LANDED	Organ	0 u
For-each (R11) after R8	wf_for_each	Fork a connector per comma-list item with {item} bound per iteration -- LANDED	Organ	0 u
Versioning (R12) after R1	wf_defs_version	@version directive, WFVER stamp, WFVERDRIFT on cross-version resume -- LANDED	Organ	0 u
Certified console (R7)	head	nx_wflow_ui console CERTIFIED 1000 permil by nx_ui_judge -- LANDED	Organ	0 u
Inbound webhooks (WA1) after R1	wfh_webhook_accept	An HTTP door that authenticates a signed inbound call, dedups by delivery id, and raises the engine event; gate proves a replayed delivery fires ONCE (R1 dedup) and an unsigned call is refused and recorded	Organ	1 u
Outbound connector call (WA2) after R8	wfc_connector_call	A connector step that calls an external HTTPS API over the sovereign TLS stack with a declared request shape, captures the response into run vars (R8), and retries under the engine's own policy; gate proves a 5xx is retried then routed on-failure (R10) and a 2xx body lands in a var	Organ	1 u
Canvas emitter (WA3) after R7,R12	wfu_canvas_emit	The console emits the workflow definition as an editable node canvas (nodes = steps, edges = deps and conditions) from the SAME definition file the engine runs; gate proves canvas round-trips to a byte-identical definition	Organ	2 u
Inline test and edit-in-place (WA4) after WA3	wfu_inline_test	Run one step from the canvas against a recorded event and show its captured vars; edit a step arg in place and verify the draft without firing (the CLI verify door R12 already exists); gate proves a test run writes a ledger record tagged trial and never advances the live run	Organ	1.5 u
Onboarding, docs, accessibility (WA5) after WA3	wfu_onboarding	First-run guide, per-step help drawn from the organ headers, WCAG contrast on every view; measured by nx_ui_judge on the same five axes that certified R7	Organ	1.5 u
NL-to-flow copilot (WA6) after R12,WA3	wfl_nl_to_flow	The sovereign model drafts a definition from a sentence; the draft is VERIFIED (R12 verify) before it is shown and a refused draft shows the refusal; gate proves a known prompt yields a definition that passes verify and an impossible prompt yields a refusal, never a half-flow	Organ	3 u
RPA UI step (WA7) after R8	wfr_ui_step	A step that drives a desktop or browser surface through the estate's own input-abstraction layer (nx_input_abstract) with screenshot evidence per action; gate proves a scripted form fill lands and the evidence frames are in the ledger	Organ	3 u
Record and replay (WA8) after WA7	wfrec_capture	Capture the operator's actions into a definition the engine replays (the PAD recorder category); gate proves a recorded sequence replays to the same end state via WA7 and a divergence is recorded not hidden	Organ	3 u

Milestones

Milestone	Rungs	Cumulative
M1 · Integration doors open	WA1,WA2	2 u
M2 · Visual product	WA3,WA4,WA5	7 u
M3 · Reach parity with the desktop class	WA6,WA7,WA8	16 u

Rung	Closes with	Definition of done (pre-declared)	Executor	Est.
OSWorld-Verified run of the sovereign RPA step (WB1) after WA7	wfr_osworld_run	DONE-RULE PRE-DECLARED: the OSWorld-Verified 361-task set runs on an isolated Ubuntu oracle host outside the build path, driven by the WA7 RPA step through the sovereign workflow engine and graded by the benchmark's own checkers; the receipt carries n, max steps, the success rate and the host hash, and is published beside the sheet's leaders [osworldsheet26] [osworld26]; a run with fewer than 361 tasks prints PARTIAL and never the headline	Organ	4 u
WebArena environments through the sovereign browser (WB2) after WB1	wfr_webarena_run	the self-hosted WebArena sites stand up on the oracle host and the sovereign browser drives the tasks through the WA7 step; graded by the benchmark's checkers; the receipt names the site versions [webarena23]	Organ	3 u
tau2-bench domains through the connector catalog (WB3) after R3	wfc_tau2_run	the tau2 domains and the March-2026 voice and knowledge extensions run against the benchmark's user simulator on the oracle host with the R3 connector catalog as the tool surface; pass-hat-k is published with k named [tau2bench26]	Organ	3 u
GAIA questions through the sovereign research runner (WB4) after R3	wfr_gaia_run	GAIA level 1 to 3 questions answered by the research runner with the workflow engine's tools, scored by exact match against the validation answers mirrored with licence [gaia23]	Organ	3 u
TheAgentCompany tasks (WB5) after WB1,WB3	wfr_tac_run	the self-hosted company (planning, chat, code hosting) stands up on the oracle host from the mirrored setup and the sovereign agent works its tasks through WA7 and R3; the receipt publishes the task completion score with partial credit the way the benchmark defines it [theagentcompany24]; OpenAgents' three-agent split (data, plugins, web) is the design reference for the tool surface [openagents23]	Organ	4 u

Ladder to SOTA

Two dated targets, and every rung's role toward them. Best in class is the proven, deployed leader today; the frontier is the research edge as of its month. Each target is dated by a sotabar row and judged against the current month by the bar-age ruler (FRESH, ATTESTED with the fallback month named, or refused). Each rung is substrate (what the target's own methods consume), an arm (a sovereign arm that triangulates without competing for the number), a contender (its done-rule is a number against the target) or superseded (kept as capability, retired as a claim), and carries the rungs it stands on.

best-in-class (T-BIC-OSWORLD), dated by SB-OSWORLD = 2026-08 ATTESTED. Intelligence-Indeed Agent 90.19 percent at 100 steps on the 361 OSWorld-Verified tasks (2026-07-25), with claude-fable-5 at 85.96 (2026-08-01) the newest dated leader on the sheet; the estate's WB1 run publishes its own success rate, n, max steps and host hash beside these ref osworldsheet26

frontier (T-FR-TAU2), dated by SB-TAU2 = 2026-07 ATTESTED. pass-hat-k on the tau2-bench domains plus the March-2026 voice and knowledge extensions under the v1.0.1 grading of July 2026, the newest published grading of tool agents in dual-control settings; the leaderboard number itself is rendered client-side and is not yet read from mirrored bytes; the estate's WB3 run publishes pass-hat-k with k named ref tau2bench26

best-in-class (T-BIC-WEBARENA), dated by SB-WEBARENA = 2024-12 ATTESTED. the top row of the WebArena public leaderboard sheet under its newest dated update (2024-12-20), attested this month as unchanged; the sheet is a Google document not yet mirrored, so the number is stated as the sheet's top row until it is; the estate's WB2 run publishes its task success rate with the site versions named ref webarena23

best-in-class (T-BIC-GAIA), dated by SB-GAIA = 2023-11 UNATTESTED. the top of the GAIA leaderboard on the level 1 to 3 validation split by exact match, under the 2023-11-21 paper attested this month as the newest version; the leaderboard is a client-side space not yet mirrored, so the number is stated as its top row until it is; the estate's WB4 run publishes exact-match accuracy per level ref gaia23

best-in-class (T-BIC-TAC), dated by SB-TAC = 2024-12 ATTESTED. the top task-completion score with partial credit on TheAgentCompany's self-hosted company (setup dated 2024-12-08, attested this month as the newest); the paper's own table is the number once its mirrored bytes are read; the estate's WB5 run publishes the same score the way the benchmark defines it ref theagentcompany24

Rung	Role	Toward	Stands on	Why
R1	substrate	T-BIC-OSWORLD	-	the durable event-sourced run ledger every benchmark run is journaled on
R3	substrate	T-FR-TAU2	R1	the fail-closed connector catalog is the tool surface the tau2 domains and the research runner call through
R4	substrate	T-BIC-OSWORLD	R1	approval steps are how a consequential desktop action parks for a human decision
R8	substrate	T-BIC-OSWORLD	R1	variables and substitution bind each task's inputs into the step argv
R9	substrate	T-BIC-OSWORLD	R8	per-step conditions are how a run branches on the checker's verdict
R11	substrate	T-BIC-OSWORLD	R8	for-each is how a 361-task set fans out over the engine
R12	substrate	T-BIC-OSWORLD	R1	versioning stamps the definition every published run was produced by
R7	substrate	T-BIC-OSWORLD	-	the certified console is the operator surface the runs are watched from
WA1	substrate	T-BIC-TAC	R1	inbound webhooks are how the self-hosted company's services call the agent back
WA2	substrate	T-FR-TAU2	R8	the outbound connector call is the step a tool-agent domain is driven with
WA3	substrate	T-BIC-OSWORLD	R7,R12	the canvas emitter renders the flow a benchmark run executes
WA4	substrate	T-BIC-OSWORLD	WA3	inline test is how one step is trialled before a full run
WA5	substrate	T-BIC-OSWORLD	WA3	onboarding, docs and accessibility are the console's own judged axes, kept beside the benchmark axis
WA6	arm	T-BIC-OSWORLD	R12,WA3	the sovereign NL-to-flow copilot drafts the plan a run follows; it triangulates planning quality without publishing the success rate itself
WA7	substrate	T-BIC-OSWORLD	R8	the RPA UI step is the desktop driver WB1, WB2 and WB5 execute through
WA8	substrate	T-BIC-OSWORLD	WA7	record and replay supplies the evidence frames every graded run carries
WB1	contender	T-BIC-OSWORLD	WA7	publishes the estate's OSWorld-Verified success rate over 361 tasks against the 90.19 leader
WB2	contender	T-BIC-WEBARENA	WB1	publishes the estate's WebArena task success rate with the site versions named
WB3	contender	T-FR-TAU2	R3	publishes pass-hat-k with k named on the tau2 domains and extensions
WB4	contender	T-BIC-GAIA	R3	publishes exact-match accuracy per GAIA level against the mirrored validation answers
WB5	contender	T-BIC-TAC	WB1,WB3	publishes the task completion score with partial credit on the self-hosted company

Ladder verdict. LADDERED. Targets, rung roles and bars are counted in the tables below, each partition re-derived beside its total. Bars read STALE against the current month 2026-10.

Ladder verdict, counted (17 rows, more than the 12-row disclosure bar)

Ladder verdict, counted
Class	Count	Note
targets	5	sotatarget rows on the plan
best in class	4	targets whose class is best-in-class, by code
frontier	1	targets whose class is frontier, by code
undated	0	targets naming no bar that exists
bad class	0	targets whose class is neither word
duplicate	0	target ids an earlier row already carries
rungs	21	rung rows on the plan
placed	21	rungs carrying a rungrole row
unplaced	0	rungs with no role toward any target
bad role	0	rungrole rows naming a role the ladder does not know
bad target	0	rungrole rows naming a target no sotatarget row declares
rungs, summed	21	RECONCILED: the parts sum to the total 21
substrate	15	what the target's own methods consume
arm	1	a sovereign arm that triangulates without competing for the number
contender	5	its done-rule is a number against the target
superseded	0	kept as capability, retired as a claim
placed, summed	21	RECONCILED: the parts sum to the total 21

Bar freshness, counted
Class	Count	Note
bars	6	sotabar rows on the plan
fresh	0	the bar month is the current month
attested	5	an older bar with a this-month barscan reporting no newer entry
stale	0	a this-month scan saw a newer month the bar ignores
unattested	1	nobody looked this month
malformed	0	a sotabar row missing a field
duplicate	0	bar ids an earlier row already carries
bars, summed	6	RECONCILED: the parts sum to the total 6
current month	2026-10	the month every bar is judged against

Treatment verdict. NOT DECLARED. No rung on this board declares its three-step treatment (a treat row naming the MCP row, the API route, the primitive it composes, the workflow that composes it and the agent beat that runs it); the ranker labels every rung TREATMENT-UNDECLARED until it does.

Inspect a rung and its prerequisites

Declared nodes 21. Rank input binding: STALE_OR_INVALID_BINDING. Dependency order is authored. Implementation, acceptance evidence, authority and resource readiness are unverified. No action is recommended or dispatched here.

Plan SHA-256 990a761daea0b683c2bd539fd3a70e7dc8e3f4659e27a0d79cf003138dc534f4. Target rows 5; role rows 21. Existing risks and release worklog retain their own scope; no node completion is inferred.

Runtime observation coverage: 0 of 21 rungs have current, identity-bound execution observations. Observation may report RED. Accepted completion remains unknown; source presence, qualification and deployment do not establish it.

Use Enter or Space on a rung to inspect its contract. Prerequisite links locate another rung in this list; open its summary to inspect it. Estimates are authored effort, not forecasts.

R1 — Durable workflow engine
Prerequisites: None declared; this does not establish execution eligibility.
Contract: wf_resume
Acceptance: Event-sourced run ledger, per-step retry, durable dedup, crash-resume -- LANDED gated GREEN
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
R3 — Organ connector catalog
Prerequisites: R1 (acceptance unverified)
Contract: wf_exec_organ
Acceptance: Exec-organ over a fail-closed catalog -- LANDED gated GREEN
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
R4 — Approval steps
Prerequisites: R1 (acceptance unverified)
Contract: wf_decide
Acceptance: Park until an APPROVED or DENIED ledger decision lands, deny wins -- LANDED
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
R8 — Variables and substitution
Prerequisites: R1 (acceptance unverified)
Contract: wf_subst
Acceptance: Event fields bind as run vars, connector stdout captured, placeholder substitution incl forked argv -- LANDED
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
R9 — Conditions
Prerequisites: R8 (acceptance unverified)
Contract: wf_cond_ok
Acceptance: Per-step k=v condition vs the variable plane, mismatch SKIPs recorded -- LANDED
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
R11 — For-each
Prerequisites: R8 (acceptance unverified)
Contract: wf_for_each
Acceptance: Fork a connector per comma-list item with {item} bound per iteration -- LANDED
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
R12 — Versioning
Prerequisites: R1 (acceptance unverified)
Contract: wf_defs_version
Acceptance: @version directive, WFVER stamp, WFVERDRIFT on cross-version resume -- LANDED
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
R7 — Certified console
Prerequisites: None declared; this does not establish execution eligibility.
Contract: head
Acceptance: nx_wflow_ui console CERTIFIED 1000 permil by nx_ui_judge -- LANDED
Authored effort: 0. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA1 — Inbound webhooks
Prerequisites: R1 (acceptance unverified)
Contract: wfh_webhook_accept
Acceptance: An HTTP door that authenticates a signed inbound call, dedups by delivery id, and raises the engine event; gate proves a replayed delivery fires ONCE (R1 dedup) and an unsigned call is refused and recorded
Authored effort: 1. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA2 — Outbound connector call
Prerequisites: R8 (acceptance unverified)
Contract: wfc_connector_call
Acceptance: A connector step that calls an external HTTPS API over the sovereign TLS stack with a declared request shape, captures the response into run vars (R8), and retries under the engine's own policy; gate proves a 5xx is retried then routed on-failure (R10) and a 2xx body lands in a var
Authored effort: 1. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA3 — Canvas emitter
Prerequisites: R7 (acceptance unverified), R12 (acceptance unverified)
Contract: wfu_canvas_emit
Acceptance: The console emits the workflow definition as an editable node canvas (nodes = steps, edges = deps and conditions) from the SAME definition file the engine runs; gate proves canvas round-trips to a byte-identical definition
Authored effort: 2. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA4 — Inline test and edit-in-place
Prerequisites: WA3 (acceptance unverified)
Contract: wfu_inline_test
Acceptance: Run one step from the canvas against a recorded event and show its captured vars; edit a step arg in place and verify the draft without firing (the CLI verify door R12 already exists); gate proves a test run writes a ledger record tagged trial and never advances the live run
Authored effort: 1.5. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA5 — Onboarding, docs, accessibility
Prerequisites: WA3 (acceptance unverified)
Contract: wfu_onboarding
Acceptance: First-run guide, per-step help drawn from the organ headers, WCAG contrast on every view; measured by nx_ui_judge on the same five axes that certified R7
Authored effort: 1.5. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA6 — NL-to-flow copilot
Prerequisites: R12 (acceptance unverified), WA3 (acceptance unverified)
Contract: wfl_nl_to_flow
Acceptance: The sovereign model drafts a definition from a sentence; the draft is VERIFIED (R12 verify) before it is shown and a refused draft shows the refusal; gate proves a known prompt yields a definition that passes verify and an impossible prompt yields a refusal, never a half-flow
Authored effort: 3. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA7 — RPA UI step
Prerequisites: R8 (acceptance unverified)
Contract: wfr_ui_step
Acceptance: A step that drives a desktop or browser surface through the estate's own input-abstraction layer (nx_input_abstract) with screenshot evidence per action; gate proves a scripted form fill lands and the evidence frames are in the ledger
Authored effort: 3. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WA8 — Record and replay
Prerequisites: WA7 (acceptance unverified)
Contract: wfrec_capture
Acceptance: Capture the operator's actions into a definition the engine replays (the PAD recorder category); gate proves a recorded sequence replays to the same end state via WA7 and a divergence is recorded not hidden
Authored effort: 3. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WB1 — OSWorld-Verified run of the sovereign RPA step
Prerequisites: WA7 (acceptance unverified)
Contract: wfr_osworld_run
Acceptance: DONE-RULE PRE-DECLARED: the OSWorld-Verified 361-task set runs on an isolated Ubuntu oracle host outside the build path, driven by the WA7 RPA step through the sovereign workflow engine and graded by the benchmark's own checkers; the receipt carries n, max steps, the success rate and the host hash, and is published beside the sheet's leaders [@osworldsheet26] [@osworld26]; a run with fewer than 361 tasks prints PARTIAL and never the headline
Authored effort: 4. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WB2 — WebArena environments through the sovereign browser
Prerequisites: WB1 (acceptance unverified)
Contract: wfr_webarena_run
Acceptance: the self-hosted WebArena sites stand up on the oracle host and the sovereign browser drives the tasks through the WA7 step; graded by the benchmark's checkers; the receipt names the site versions [@webarena23]
Authored effort: 3. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WB3 — tau2-bench domains through the connector catalog
Prerequisites: R3 (acceptance unverified)
Contract: wfc_tau2_run
Acceptance: the tau2 domains and the March-2026 voice and knowledge extensions run against the benchmark's user simulator on the oracle host with the R3 connector catalog as the tool surface; pass-hat-k is published with k named [@tau2bench26]
Authored effort: 3. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WB4 — GAIA questions through the sovereign research runner
Prerequisites: R3 (acceptance unverified)
Contract: wfr_gaia_run
Acceptance: GAIA level 1 to 3 questions answered by the research runner with the workflow engine's tools, scored by exact match against the validation answers mirrored with licence [@gaia23]
Authored effort: 3. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.
WB5 — TheAgentCompany tasks
Prerequisites: WB1 (acceptance unverified), WB3 (acceptance unverified)
Contract: wfr_tac_run
Acceptance: the self-hosted company (planning, chat, code hosting) stands up on the oracle host from the mirrored setup and the sovereign agent works its tasks through WA7 and R3; the receipt publishes the task completion score with partial credit the way the benchmark defines it [@theagentcompany24]; OpenAgents' three-agent split (data, plugins, web) is the design reference for the tool surface [@openagents23]
Authored effort: 4. Executor kind: Organ. Responsible, accountable and verifier not established. Inspect retained worklog.
Runtime observation: UNKNOWN; outcome UNKNOWN; observed epoch -1. Declared owner: . Receipt , SHA-256 . Identity-bound observation is not accepted completion. Qualification and deployment remain unverified by this projection.

Learning and practice paths

No structured learning path is declared for this plan. Existing research, roadmap and worklog remain available above.

Rung dossiers — every record that bears on a rung, in one place

Joined by symbol, never typed. Each dossier is assembled on every publish from the rows this board already carries: the plan's rung row, the matrix row that carries its contract symbol (its status re-measured by the one symbol ruler), the gate the organ declares and the verdict the evidence referee recorded on its last run, the reference marks in the rung text, the design rows that name the rung and the work log filed against it. The capability matrix links each cell to its dossier and each dossier links back to its cell and forward to the rungs it stands on; a dependency naming no rung on this plan is written as UNRESOLVED rather than linked. Absences are named: NO-MATRIX-ROW, NO-GATE-DECLARED, UNRECORDED.

Gate verdicts: UNSTAMPED — no per-gate verdict artifact is published for this domain yet (nx_swcompare_evidence automation writes it beside its stamp); every gate below reads UNRECORDED until it runs.

Gate declarations: knowledge/organ_gate.conf read from this working directory (1 rows) and from the estate root (196 rows); rows in both counted once: 0; declarations read: 197. Every declaration in either copy counts, so a stub beside the publisher can never hide the root.

R1 Durable workflow engine · cell PRESENT · gate NO-GATE-DECLARED

Contract	`wf_resume` · executor Organ · 0 u · stands on nothing
Definition of done	Event-sourced run ledger, per-step retry, durable dedup, crash-resume -- LANDED gated GREEN
Matrix cell	Unified multi-step durable workflow engine · `runtime/nx_wflow_engine.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_engine`; declare `nx_wflow_engine TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	2026-09-02 measure The domain's executed proof (automation.gates: nx_sclass_dispatch_gate) read RED 1/4 on 2026-09-02 while the box was saturated: its proof-gate build was REFUSED by admission, and the gate convicted the subject. Two defects, both in the gate: it hand-tallied pass/fail into a one-state gv_ctr (a dropped conjunct is invisible), and it matched the runner's PRINTED rc=4 where nx_sov_build_run exits SBR_ADMIT_REFUSED=6 for every denial (3/4/5 collide with COMPILE_FAIL/ASM_FAIL, by design). Fixed: per-tooth gv_check, gv_need SKIP on exit 6 (subject unexamined, never RED), T3 liar-kill runs regardless; rebuilt 97,515 B sha d1d0eefe, promote in flight. The referee itself was also fixed the same day (private capture path per run; exit 124 is the referee's deadline, not the gate's verdict). 2026-09-02 land nx_sclass_dispatch_gate PROMOTED 97515 B sha d1d0eefe (gv_check per tooth, gv_need SKIP on runner exit 6, T3 phantom-gate liar-kill runs regardless) and the referee nx_swcompare_evidence PROMOTED 115389 B sha 57975459 (per-run private capture path; a deadline kill before any verdict line is DEADLINE-UNMEASURED = abstain). Re-refereed 2026-09-02: dispatch gate SKIP exit=3 pass=1/1 under a REFUSED builder admission, subject unexamined; evstamp_automation grounded=26/26 gates=0/1 skipped=1 redseen=0 unproven=1 unproven_reason=every-gate-abstained. The domain no longer convicts itself on a saturated box: it abstains until the builder admits, and reads UNPROVEN not RED on /compare and in nx_deploy_ready. 2026-09-13 measure 2026-09-13 native workflow evidence repair adopted and callable: nx_wflow serving and offc artifacts 110128 B SHA256 37b380449305b5fdd888faf6308a291b1903985bc512e2b9c88371ff3bfbc0c0 independently verified. Required transition journal failures stop execution; canonical gate nine failure checks plus legacy state selftest passed. Exact post-action receipt fault fixture passed 12 checks. Authored MCP schema adopted; scoped discovery reconcile first registered=1 then already=1. Actual admin API fire completed harmless variable plus retry-probe workflow; durable ledger retained, repeat event deduped unchanged, directory-ledger negative refused before action. No message or task stubs count as delivery. No public or wildcard grant change. Alpha-Beta-Production product promotion, rights authority, crash reconciliation and direct attached MCP credential validation remain unverified; directory read refusal has zero diagnostic output. Full evidence knowledge/gates/workflow-exposure-candidate-t230.json; API jobs1789309701,1789309711,1789309720. No claim of exactly-once or photoreal improvement. 2026-09-19 audit 2026-09-19 dynamic identity repair pending: native gate1789860000 proved distinct long event IDs alias under canonical wf_evid truncation; 27/27 containment/result tests do not close full identity. nx_wflow_engine wf_fire, wf_resume, wf_decision, wf_emit and observer formatters retain fixed identity buffers. Reuse nx_wflow_text_owned measured allocation; reviewed versioned lossless new RID plus evidence-qualified legacy lookup must preserve existing journal bytes and refuse uncertain historical binding. No new engine adoption yet. 2026-09-20 verified 2026-09-19 dynamic workflow identity/result closure now canonical and serving: promotion1789864076 live/offc197664B SHA4c570b50b807d6bbb96f05cba273139443b0eb9c93d78ae7b39853d352ecf012, identity44/44 and scoped ownership6/6. Original event-result1789864130 read-only response1357B, historical ledger unchanged6602B; command_execution_reaped0 and served_acceptance unverified explicitly retained. Full evidence buildroot/knowledge/gates/alpha-release-workflow-delta-t343.json. This supersedes candidate status above, not history. 2026-09-20 debt 2026-09-19 native boolean boundary evidence: probe1789863739 evaluates RHS once for both true OR probe and false AND probe. Wrapper1789863616 crashed after assuming short-circuit argv guard; explicit nested argc branches qualify44/44 in1789863672. No compiler bug asserted: shared review/lint must flag guard-and-dereference assumptions under measured language semantics. Evidence buildroot/knowledge/gates/workflow-eager-boolean-evidence-t344.log and workflow-argv-boundary-failure-t344.log. No estate-wide lint coverage claimed. 2026-09-20 debt 2026-09-19 repeated allocator-contract defect: shipping behavior candidate t371 initially used aborting sys_mmap followed by null checks for argv/path scratches (source delta cb7d5094, compile7693); those structured refusal branches were unreachable on allocation failure. Prior workflow ownership review independently established sys_mmap_try/direct-unmap contract. Corrected current owner nx_organ_ship_review_candidate_t371 uses one checked argc/path extent and a sys_mmap_try cleanup wrapper through every inner return; compile7807 actual usage exit2 verified, allocation/workload gate still pending. Existing shared source-review/lint capability must detect aborting allocation followed by recovery checks and distinguish arena from direct ownership; no estate-wide detector coverage or leak-freedom claim. Exact repair/evidence: buildroot/knowledge/gates/shipping-behavior-review-t371.json. 2026-09-20 debt 2026-09-19 candidate-only authority conflation caught before adoption: t371 renamed loss_approved to live_recheck_required, then inherited use also sent allow_capability_loss=yes for every successful no-loss or behavior-reviewed adoption. Root source review rejected this; new exact_loss_approved output is set only by successful os_loss_exact, separately from live identity recheck. Existing shared boundary review/lint must flag verification/observation states reused as authorization to weaken another gate. No deployed authority expansion claimed. Actual promotion-body countertests and full correction retained in buildroot/knowledge/gates/shipping-behavior-review-t371.json. 2026-09-29 land 2026-09-29 ENABLEMENT: native owned PM transport installed and discoverable as nx_pm_owned_native. Exact serving/offc ELF 557321 bytes SHA256 b89678b182fee20e81160f23357e38cac294c5b20eed0a852e3640391e550996; native NTH1 host policy 594 bytes SHA256 7f4d75585bf99bb9ea2ae3523033228c4f8cf302f3593694c423954226a692cb. Existing native approval/input/version/claim owners enforce current signed operator identity, scoped capability, registry, RACI, exact version approval and replay. Private qualification job1790661374 passed43/43 with real native worker, swaps, role/approval withdrawal and scoped fixture cleanup. Install1865, register1938 and one-row discovery2070 completed; actual served invalid-intent probe1982 returned structured400/exit2 with execution_attempted=0, an expected negative test. This is installed transport, not an adopted edit-build workflow, autonomous repair, net resource savings or Beta/production acceptance. Positive current-host credential invocation and end-to-end owner workflow adoption remain open; Claude hook candidate remains uninstalled. Retained worklog buildroot/runtime/clmem/project-owned-alpha-native-architecture-2026-09-28.md. Browser connection discovery returned no available browser at2026-09-29T06:08Z; visual release review remains pending, not waived. 2026-09-29 land 2026-09-29 ENABLEMENT: native PM credential handoff installed and callable as nx_pm_native_handoff, ELF255547 bytes SHA2566fd8868ddf94135bb4a0ae1aa02f81458ed0c60a7dfbbbc92ea9b40fff2acd09. Private25/25 gate1790665106 verified real signed identity, exact workflow capability, current roles/revocations, complete input, safe replay, private output and cleanup28 files/324595 bytes/remaining0. Actual host5163 retained351-byte native NAC1 without executing a workflow; installed API5394 and direct MCP5459 reconciled the same artifact without replacement or secret output. JSON is used only for management/MCP interoperability; native output and existing NTH1 authority remain. Registered5294; scoped discovery5395 rows1/failed0. Existing workflow binding was not migrated. The complete owned edit-build-qualify-apply-verify loop, typed command semantics and automatic credential retirement remain open; no autonomous repair, net savings, Beta or frontier claim. Detailed retained report knowledge/gates/native-handoff-adoption-20260929.md; architecture worklog project-owned-alpha-native-architecture-2026-09-28. Browser recheck07:05Z returned none; visual release review remains open. 2026-09-29 land 2026-09-29 ENABLEMENT: native PM staging and approved selection installed as ELF0a22b25139b0f20705769c6419bf3022f5dec678a90dda80e3179491d5e020c4 (585529 bytes), exact build/root/_offc identity verified. STAGE retains native candidates without approval or selection; SELECT requires current exact approval and compare-and-swap; admitted requests keep their implementation and rollback replay cannot reapply an old transition. Private35/35 plus transport43/43 passed. Actual MCP8145 verified installed input refusal after owner-led renewal of an expired entry credential; no automatic healing claim. Real native artifact verifier separately passed25/25 with signed stage/approve/select/admit/invoke and retained match/mismatch/unavailable outcomes; current-host workflow adoption is still OPEN and verifier is not installed. Qualification cleanup330 private files/3076771 bytes/remaining0; no net estate savings claim. Report knowledge/gates/native-stage-select-adoption-20260929.md. Native authoring API, canonical input budget admission, full owned repair workflow, credential retirement and browser visual review remain open; browser inventory empty07:50Z. Alpha; no Beta/frontier/autonomous repair claim. 2026-09-29 land 2026-09-29 ENABLEMENT: native authoring installed in nx_pm_owned_define (792636 bytes SHA2561359391b8acf8aeb49769b3522900798dcbe6f70d8ae5b29f6d7326b1c13d92c), qualified29/29. Existing Nishi owners now author native versions, six-action intents and authority-preserving host budget proposals. Current-host estate-native-control-verification-v1 completed signed stage/approve/select/admit/invoke: three installed artifacts matched; exact replay reused receipt af6f17bc without execution. Native policy read budget262144 applied with exact forward and retained reverse intent. Existing discovery reconciler gained scoped refresh-v1, qualified20/20 and installed a01e1bff; live refresh changed one description and repeat held generation4142, with registration/index unchanged. Actual served MCP contracts verified09:01Z. Original private test failures retained and corrected; qualification fixtures removed with verified absence. Detailed report knowledge/gates/native-authoring-adoption-20260929.md and retained architecture worklog. Full owned repair/build/qualify/install/verify, general native receipt inspection, credential retirement, legacy bulk discovery limits and visual release review remain OPEN. Browser inventory empty08:57Z; served bytes do not establish visual acceptance. Alpha; no autonomous recovery, Beta/frontier or net savings claim. 2026-09-30 land 2026-09-30 ENABLEMENT: native PM retained-request inspection is installed and exercised through the existing management API. INSPECT-only adapter702088B SHA826e3340031289b4f9750bb93827afbb54a9170f01352593f82d7a887be540ef and shared transport702104B SHAc5034f4821207c7e16c706a26a62dcb6c5e7d5bcd9ed718043152b051a82213b have verified root/offc equality. Current private inspection gate54/54 rejects all six effect actions at the read-only entry even with execution credentials; separate read scope, original actor, budgets, version retention, failure/expected-nonzero distinction and no-rerun behavior are verified. Installed author, scope registration, private grant mint and handoff drove actual inspection1790782421 and repeat1790782664: exact historical receipt af6f17bc retained, claim/generation unchanged, worker execution0, read1916005B/21 files. This is historical Sept29 ARTIFACTS_MATCH evidence, not a fresh current-artifact check. Native NTH2 explicit budgets installed with reverse intent; handoff/definition updates installed and qualified. New definition-gate cleanup45/45 includes foreign-file refusal and repeat-safe own-fixture removal. Discovery contracts and index membership verified; provider caches not claimed refreshed. Detailed report knowledge/gates/native-receipt-inspection-scope-20260929.md; retained architecture worklog project-owned-alpha-native-architecture-2026-09-28. Open: complete owned edit-build-qualify-apply-verify/close workflow, automatic credential retirement, long-lived allocation accounting, broader consumer adoption, original incident closure and comparative resource return. Browser verification through registered NAS capture1790782866 reported CD-BROWSER-ABSENT: configured candidates target Windows/WSL, no resolved browser on executing host. Visual/interaction review remains OPEN; served bytes do not close it. Alpha; no self-healing, estate-wide SOTA, Beta or net savings claim. 2026-09-30 land 2026-09-30 ENABLEMENT: the existing Nishi shipping owner now authors and consumes native NGP1 per-gate inputs. Installed nx_organ_ship329176B SHA31f1ea03162326481b19e78db3bda563cb3cdba70941dbb65efeccbd03143295 has independently matched serving/runner copies. New gate62/62 and existing71/71 plus151/151 passed. Installed author replay5249 retained the same1368B native recipe172a5440 with no duplicate; real consumer build1790785287 used13 inspection inputs and8 definition inputs separately, passed54 plus41+4 checks, and promoted nx_pm_owned_define884592B SHAb44f48328efdc001bc7cbb993bec96083c0223bb5916cf49c60efebef3082b5b with independent serving/runner equality. Exact input/config identities, required-gate coverage, literal arguments and per-execution ownership are enforced; malformed, conflicting and over-budget records refuse. Source-closure banking6066 produced byte-identical qualified native gate79928f66; current closure independently checked6384. The definition gate now emits its canonical verdict after all cleanup, qualified47/47 in6334 and accepted by the original build API6383 as375600B SHAa803cfd5. Its100 isolated test files/21686B were removed and directory absence independently confirmed. These gate artifacts are qualification-only; no separate gate installation claimed. Discovery row refreshed4158; provider caches unverified. Detailed retained worklog buildroot/runtime/clmem/project-owned-alpha-native-architecture-2026-09-28.md names original evidence, rollback copies, superseded candidates and fixture debt. OPEN: automatic permanent gate workloads with legacy-call compatibility, old shipping-gate fixture disposition, full PM-owned repair/build/verify/recover/close, credential retirement, original incident closure and comparative resource return. Whole-conf recipe invalidation remains strict. Browser execution-host/config integration and served visual/interaction review remain OPEN. This is one installed and exercised build-composition capability in alpha, not autonomous recovery, estate-wide SOTA, Beta or measured net savings. 2026-10-01 measure BARFRESH 2026-10-01 (seat residue-fleet, T0 1790856940): before STALE with 6 of 6 bars UNATTESTED on the October clock; after 5 ATTESTED and 1 UNATTESTED, verdict still STALE. Attested from bytes re-fetched over the sovereign fetcher and read this month, all five byte-identical to their 16 Sep mirrors: SB-OSWORLD (results sheet re-read through nx_xlsx, 1077 rows, largest Date serial 46235 which is 2026-08-01), SB-TAU2 (README, newest dated item July 2026), SB-WEBARENA (README, newest 12/20/2024), SB-TAC (README, newest tag 20241208) and SB-OPENAGENTS (README, newest news 2023 Oct 26). Left for the arXiv attest runner and NOT fetched by this seat: SB-GAIA (gaia23, an arxiv.org abstract page). The page needs a regen to show the October barscans.

R3 Organ connector catalog · cell PRESENT · gate NO-GATE-DECLARED

Contract	`wf_exec_organ` · executor Organ · 0 u · stands on R1
Definition of done	Exec-organ over a fail-closed catalog -- LANDED gated GREEN
Matrix cell	Sovereign organ-connector catalog (any organ as a step) · `runtime/nx_wflow_engine.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_engine`; declare `nx_wflow_engine TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	2026-09-13 audit 2026-09-13 CORRECTION: prior log1789312276 used GE57, which is not a rung in this automation plan; source-discovery evidence belongs to the organ connector catalog R3. Prior entry retained as history. Current attached MCP still returned admitted_workers_in_use dispatched=false during independent reads; serial retry succeeded. Fresh-config bridge qualification does not establish attached transport replacement. No concurrency recovery or estate-wide reliability completion claimed. 2026-09-13 audit 2026-09-13 SWARM EXECUTION CONTRACT: retained verified workstream knowledge/gates/beach-asset-subscriptions-workstream-t251.json includes verified tool contracts, access, representative preflight, existing owner repair scope, input/output/acceptance and review evidence. Contract recorded; automated enforcement not yet delivered. Actual Claude control probes established enforced available-model list prevents alternate model selection without inference. Existing adapter candidate now waits for same-child initialization and fresh billing/account evidence before user message; root reviewed full worker and requested lifecycle/protocol negative tests. No provider task launched or tokenROI claimed. Public source observations retained asset-source-observations-t253.json; first task is supplied-evidence analysis, not claimed live browsing. Shared fetch classifier native regression moved1/3 to3/3 job1789314037; broader and live validation pending. Beach inspection integration proceeds separately in existing shared owner. 2026-09-13 review 2026-09-13 Root reviewed full Claude preflight candidate4a711b52c3c4a9fb7fc359a86d59af4368d47436e468d6a023ae6a555d0f36e5, exact diff and qualification21/21 protocol plus4/4 receipts. Approved first public supplied-evidence unit through existing host facade over MCP, actualaster/Fable5.1, toolsdisabled, no paidfallback. No claim canonicalWindowsadoption/native registration or automatedexecutionBriefvalidation. Actual provider result stillpending. Correction to prior livefetch failure interpretation: witness caller1MiB outputcapacity exhausted, underlyingdeflateoverflow; generic-33 mapping obscured it. Not evidence malformedgzip or productiondefaultregression. Worker retaining measuredcompressedpayload and repairingclassification without arbitraryhigherlimit. Root review files claude-preflight-qualification-t251.json and fetch-status-repaired-review-t252.json. 2026-09-13 requirement OPERATOR investment clarification: prioritize SOTA quality and timely delivery; optimize by growing Nishi's repeatable execution and evidence capabilities, not reducing quality, detail or required review. Evaluate combined accepted user outcome and reduced repeat coordination. Source presence or a green isolated test alone does not establish reusable delivery. Measure real workflow invocation, acceptance evidence, root supervision effort, provider usage and runtime cost separately; unknowns stay unknown. Current focus remains Beach delivery and supervised ingestion/swarm dispatch; no new parallel domain opened by this clarification. 2026-09-13 delivery 2026-09-13 Root native release-manifest published Beach Alpha inspection-distance page d83d169c1d223f12be6ab1dfdeb751872427e42ceea941d8fbbd73410c6b8647; actual Chrome served hash and zoom/Fit/gameplay restoration verified, errors0. Evidence knowledge/gates/inspection-distance-served-root-t260.json. This is diagnostic inspection delivery, not character beauty or Beta acceptance. Shared standards required: failure categories remain distinct through consumers; source/build/configured/running/served identities must not be conflated; release preparation derives missing registration from actual dependency registry and preserves hash guards. Earlier manifest existing-only registration refusal retained715caafd; exact reviewed registration via existing release workflow succeeded352d2792. Existing owners tasked to make preparation/reload verification reusable; automatic completion not claimed. No provider trials resumed. 2026-09-13 measure 2026-09-13 Shared fetch capacity repair adopted through nx_organ_ship5704 into nx_https_get_cli serving/offc SHA256 d858f100637f21244f444b1665566aa7733807cca08e965333b0bc91343e0be1 577560B; independent live hash5744 matches. Reusable rule: distinguish caller capacity exhaustion (HF_DEC_CAPACITY -34), malformed compression (-33), framing/dependency errors, and incomplete captured evidence; never treat decoded-too-large as raw-success. Native retained RigModels gzip expands306799 to2439009B with CRC and byte equality; boundary gate5219 passes13checks; live default5228 returns200 and fulldecodedbody. Actual attached MCP5743 decodedHTML but output contains explicit CAPTURE-TRUNCATED marker at1048576B; classify incomplete, not complete ingestion. Owner runtime/nx_tools_api.nx1663..1690 caps async capture and marker lacks structured completeness. No cap increased. Source fixes do not prove all deployed shared consumers rebuilt. Existing decoder MAX_BLOCKS exitwithoutfinal still falsely reportsOK: queued against runtime/nx_deflate.nx, not repaired here. Canonical webscraping/dataio Compare owner unverified; integration evidence retained here. Full review buildroot/knowledge/gates/fetch-capacity-review-t253.json. 2026-09-13 measure 2026-09-13 Character source-descriptor/rig-preservation operation nx_gltf_render_prepare registered5826 through existing API alias to existing nx_gltf_render_prepare_candidate_t230.elf; no duplicate implementation. Guarded promotion5795 self-gated exactfreshartifact with retainedCatwalk GLB, serving/offc SHA2565ea8ce60601bc19fbc5a1aa727013bc21248bc37795ec865f57c9bea30c18c63 143320B hashes5827/5851. Actual API invocation5844 returned34024B descriptor c8f1397ce722308ec280b54ea77b4ca617aa279ab5bf48a81b22306869eef3e0 reused1. Discoveryreconcile5845 registered1 failed0; authoredschema title includes arguments and scope. Attached-client newtool exposure unverified. Initialship5737 refused because operator omitted requiredgatearguments; retained as invocation failure, not subjectdefect. This preserves existing source rig; not autorigging, appearanceimprovement, Beachintegration, or fullingestionworkflow. t247 boundaryreview retained; nestedallocation/durability limits remain. 2026-09-13 verify Public discovery now verified at https://nishifamily.com/api/tools job1789315927; retained784543B output contains exact nx_gltf_render_prepare entry with full argument/scope description near offset780043 and JSON end. Actual API invocation5844 alreadyverified. Attached direct-tool handle refresh remains separate. Full adoption record buildroot/knowledge/gates/fetch-prepare-adoption-t253.json. No Compare public regeneration claimed. 2026-09-13 verified 2026-09-13 Root independently invoked registered nx_gltf_render_prepare via native mgmt API against retained source GLB, mesh0 skin0, exact existing descriptor: job1789316130 exit0 reused1 bytes34024 SHA c8f1397ce722308ec280b54ea77b4ca617aa279ab5bf48a81b22306869eef3e0. Full discovery JSON parsed complete1946tools with exactly1 operation row; source-descriptor/rig scope explicit, no autorigging or Beach-placement claim. Poll claim metadata before requesting output; full registry output is unnecessary context expense, project exact known record from retained evidence. Reusable API delivered; attached tool refresh, integration and consumer acceptance remain separate. 2026-09-13 measure 2026-09-13 Existing native nearest-surface asset gate now adds six independently provable off-surface support-plane probes per asset, exact nearest coordinates and squared distances. Separate Paladin5015, DarkKnight5048, Toon5055, Witch5063 each45/45; resident5082 40/40. Expanded analytic gate5073 30/30 includes tilted plane, reverse-winding edge and duplicate-vertex degeneracy. Retained complete jobs and knowledge/gates/bvh-nearest-review-t284.json; asset ELF64613B93285f82bee0ad8bb9bdc4be1181bf692809b02cfb0a64372b106c6fb8fe807e. This improves reusable native evaluator independence beyond shared-kernel exhaustive agreement; no signed contact, posed motion, hair generalization, FPS or visible Alpha improvement claimed. No public Compare regeneration. Graphics domain owner not resolved in this bounded task; reusable invocation evidence retained here. 2026-09-13 blocked 2026-09-13 Writer artifact load repair native21/21 on actual1415858B BeachHTML plus prior bound-request26/26 and shared47/47. Exact size-derived allocation replaces accidental1MiB display-limit input ceiling; source36676B3a7da1ac8e8bd39c5992fa8f01d6db4cbc9f19dfca07be84509b2c80688ca06b. NOT deployed: guardedship7154 caught stale gate assertions, corrected native7275 18/18; retry7348 refused17 printable losses revealing livewriter predates bound-fingerprint source. Native current-ledger7496 confirms malformed writer row32828..33102 SHA7463b3df779b8578fde7b342a26404ed07a373a53c301faf94293aff14e799c6, unknown original keys/pairing. New strict reader would block unrelated fresh keys; regression7533 explicitly proves this. Existing explicit-epoch-design-t219 recovered for safe additive namespace plan, not silent history deletion. Complete current recovery review knowledge/gates/writer-live-ledger-recovery-review-t293.json; no live journal mutation/allowance/ship or public Compare regeneration. Reusable lesson: source qualification plus isolated tests must be joined to actual retained state compatibility before deployment. 2026-09-19 audit 2026-09-19 reusable part-to-stage delivery: knowledge/wflow/component-release-preparation.flows and .steps already bind product/component/stage and exact manifest path/SHA; nx_site_publish_lib rm2_preflight/execute/publish accepts generic consumer source/path/hash/bytes/expected_live and immutable dependency collection. Actual Alpha experimental publication8283 reconciled ledger and served hash; DONE is not visual acceptance. Remaining gap is candidate/component plus assembly artifact and manifest identity bound to stage-specific prerequisite evidence and approval; scope labels alone do not enforce lineage. v2 requires at least one dependency. No claim all component types or Beta/Production delivery qualified. 2026-09-29 land 2026-09-29 REPAIR: Claude result hook installed with qualified ELF120858 bytes SHA256058238dd6a35e5b7ffb4ae60cbda48852d99faeab94f5999317ca075868463de. A nonzero process exit no longer asserts command failure. Native qualification157/157 passed on NAS and local host; unchanged configured Claude launcher passed11 cases/112 assertions and both exact reported jobs1790654753/1790654870 with24 assertions. Those searches returned PRESENT with590 and18 matches, complete coverage. Existing estate gate also replayed those exact client projections21/21 in job1790663486. Canonical source18789e09 adopted through native immutable install intents; canonical build3119 reused byte-identical qualified artifact. Prior source/executable and explicit reverse intents retained; install replay reconciled without another replacement. Command outcome remains UNCLASSIFIED until its contract is applied; no automatic repair/retry, active-provider-session observation or estate-wide enforcement claim. Existing debt1790624360 remains open for owned workflow/typed outcome closure. Evidence knowledge/gates/truncwarn-adoption-20260929.md and existing architecture worklog. Release visual review remains open because no browser connection was available.

R4 Approval steps · cell PRESENT · gate NO-GATE-DECLARED

Contract	`wf_decide` · executor Organ · 0 u · stands on R1
Definition of done	Park until an APPROVED or DENIED ledger decision lands, deny wins -- LANDED
Matrix cell	Human-in-the-loop approval steps · `runtime/nx_wflow_engine.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_engine`; declare `nx_wflow_engine TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	2026-09-19 audit 2026-09-19 sequenced repair after identity: reuse existing wf_decision/wf_decide parked approval with exact candidate/component/assembly/manifest/stage evidence; reject stale approvals. Experimental Alpha may expose declared failed evidence; Beta integration/rights and Production acceptance require distinct policy. No new service and no uniform all-green rule.

R8 Variables and substitution · cell PRESENT · gate NO-GATE-DECLARED

Contract	`wf_subst` · executor Organ · 0 u · stands on R1
Definition of done	Event fields bind as run vars, connector stdout captured, placeholder substitution incl forked argv -- LANDED
Matrix cell	Data passing between steps (outputs to inputs, variables) · `runtime/nx_wflow_engine.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_engine`; declare `nx_wflow_engine TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

R9 Conditions · cell PRESENT · gate NO-GATE-DECLARED

Contract	`wf_cond_ok` · executor Organ · 0 u · stands on R8
Definition of done	Per-step k=v condition vs the variable plane, mismatch SKIPs recorded -- LANDED
Matrix cell	Mid-flow branching / conditional paths · `runtime/nx_wflow_engine.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_engine`; declare `nx_wflow_engine TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

R11 For-each · cell PRESENT · gate NO-GATE-DECLARED

Contract	`wf_for_each` · executor Organ · 0 u · stands on R8
Definition of done	Fork a connector per comma-list item with {item} bound per iteration -- LANDED
Matrix cell	Loops / iterate over a collection · `runtime/nx_wflow_engine.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_engine`; declare `nx_wflow_engine TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

R12 Versioning · cell PRESENT · gate NO-GATE-DECLARED

Contract	`wf_defs_version` · executor Organ · 0 u · stands on R1
Definition of done	@version directive, WFVER stamp, WFVERDRIFT on cross-version resume -- LANDED
Matrix cell	Flow versioning / draft-test-live · `runtime/nx_wflow_engine.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_engine`; declare `nx_wflow_engine TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

R7 Certified console · cell PRESENT · gate NO-GATE-DECLARED

Contract	`head` · executor Organ · 0 u · stands on nothing
Definition of done	nx_wflow_ui console CERTIFIED 1000 permil by nx_ui_judge -- LANDED
Matrix cell	U1 product visual design (vs the Zapier editor) · `runtime/nx_wflow_ui.nx` · PRESENT decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_ui`; declare `nx_wflow_ui TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WA1 Inbound webhooks · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfh_webhook_accept` · executor Organ · 1 u · stands on R1
Definition of done	An HTTP door that authenticates a signed inbound call, dedups by delivery id, and raises the engine event; gate proves a replayed delivery fires ONCE (R1 dedup) and an unsigned call is refused and recorded
Matrix cell	Inbound webhook triggers · `runtime/nx_wflow_hook.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_hook`; declare `nx_wflow_hook TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	2026-09-13 land 2026-09-13 WORKFLOW FAILURE DIAGNOSTICS LIVE: canonical runtime/nx_wflow.nx6167B adopted with exact5477B backup knowledge/gates/nx-wflow-before-diagnostics-t240.nx. Real candidate gate job1789309922 passed21/21 across failed fire/resume/board and successful existing-run dedup; guarded ship1789309976 installed SHA ec38e8f50da26dc52273ac0da9c5ee4a7d20a7e52fc18cb53912e7833f05cd98. Actual API negative1789309996 preserves exit1 and emits nishi-workflow-error/1 native_code-1, effects unverified, retry reconcile-first; earlier9720 silently returned zero output. No claim all validation failures structured, exactly-once effects, or automated stage acceptance. Gate provenance closure remains unbanked and ship reports independent direct-edge approximation; actual candidate executable explicitly supplied to gate. Outer organ deadline discovery: ma_do_organ_run supports form deadline_ms separately from builder --run-timeout-ms; two visual workers notified after observed12s terminations. Discovery/schema guidance needs this distinction retained. 2026-09-13 audit 2026-09-13 MCP QUEUE CANDIDATE ROOT REVIEW: actual helper gate1789310332 passes13/13 with real occupied worker, FIFO, queued/running duplicate identity, pre-dispatch cancellation, memory refusal, expiry and child recovery. No installed queue claim. Root read pending candidate27286B queue implementation and stdio candidate main; found bootstrap queue_bytes=workers*request_cap omits copied-body metadata/control/normalized-id charge. With one worker a valid near-request_cap line could newly fail despite idle executor. Returned for checked envelope derivation and real boundary regression before adoption; do not just increase constant. Actual stdio stream test still pending. Native compiler aggregate-control-copy fault discovered in first candidate remains separately retained debt, not closed by explicit-field copy. This is reviewed candidate evidence, not dependable swarm completion. 2026-09-13 measure 2026-09-13 MCP admission queue candidate qualified, NOT installed: final native helper0610 GREEN13/13 including full-budget cancellation with occupied worker; exact stdio ELF gate0638 GREEN10/10 including maximum8192-byte request under default queue, explicit128-byte configured queue refusal and zero wait startup rejection. Final candidate485471B SHA25607625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6. Root-found compatibility gap corrected by checked body plus normalized-identity plus actual metadata envelope, not raised constant. Structured queue state and correlated undispatched refusals retained. Existing allocator abort paths and compiler embedded aggregate defect0497 RED1/4 remain open; no live upstream success or attached bridge repair claimed. Review knowledge/gates/mcp-admission-queue-review-t230.json. Root deployment review required. 2026-09-13 measure MCP queue reviewed canonical adoption0822/0823/0824 exact backups/readbacks complete. Canonical helper0870 GREEN13/13, exact canonical stdio0921 GREEN10/10; incumbent samepool recovery0878 GREEN14/14, control0894 GREEN13/13, HTTP framing0902 PASS10, actual HTTPS404 recovery/public initialize1124 GREEN7/7. Immutable knowledge/gates/nx_mcp_stdio-queue-t230.elf485471B SHA07625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6 copied0940 and hashed0956. Current running bridge untouched. Authenticated tool roundtrip still pending verified NAS credential-file path; public initialize is not authentication proof. Complete record knowledge/gates/mcp-admission-queue-review-t230.json. 2026-09-13 measure AUTHENTICATED MCP QUEUE PROOF: immutable nx_mcp_stdio-queue-t230.elf SHA07625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6 passed native authenticated gate1470 GREEN6/6. Exact nx_fs-only grant minted via existing private-response API1344,600s lifetime, no token output. Two actual nx_fs size requests queued under default1worker then returned canonical22199B exact1 exit0/isErrorfalse; invalid-token control denied both reason2. Full positive432B and queue events independently read. Canonical source adoption already complete; record now distinguishes historical candidate limits from current open limits. Running attached bridge not switched; preserve existing credential routes and activate immutable revision through configuration owner. Review knowledge/gates/mcp-admission-queue-review-t230.json. 2026-09-13 measure LOCAL MCP QUEUE DEPLOYMENT: immutable485471B SHA07625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6 staged through NodeMCP exclusive copy/hash. Local authenticated gate GREEN6/6 existing configuredcap; actualcwd queuegate GREEN10/10. Exact two executable references activated via checked atomicrename in Codexconfig4809B SHA916f9d4883c8c2996f2d9b96f879615b5e4931627b64fde1d7abdb8535a14dca; all routes/caps/settings preserved; backupretained. Postactivation freshnativegate exec1960d2 GREEN6/6. Existing attached connector NOTprovenreloaded; do not claim threadtransport fixed. Narrow rule27 sovereignELF launch exception retained; no shelltransfer/Pythonwrapper. Review knowledge/gates/mcp-admission-queue-review-t230.json. 2026-09-13 measure Freshbridge exactconfigured nishi_admin route verified: nx_mgmt GET/api/health accepted with existing admincap, terminal185B16services0down1degraded swap999. Childexited0 onEOF; no runningconnector killed. Hostreload notcallable in currenttoolinventory; normalserverRestart/freshattachment remainsrequired before claiming attachedrepair. Record knowledge/gates/mcp-admission-queue-review-t230.json.

WA2 Outbound connector call · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfc_connector_call` · executor Organ · 1 u · stands on R8
Definition of done	A connector step that calls an external HTTPS API over the sovereign TLS stack with a declared request shape, captures the response into run vars (R8), and retries under the engine's own policy; gate proves a 5xx is retried then routed on-failure (R10) and a 2xx body lands in a var
Matrix cell	Third-party SaaS connector catalog · `runtime/nx_wflow_connect.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_connect`; declare `nx_wflow_connect TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WA3 Canvas emitter · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfu_canvas_emit` · executor Organ · 2 u · stands on R7, R12
Definition of done	The console emits the workflow definition as an editable node canvas (nodes = steps, edges = deps and conditions) from the SAME definition file the engine runs; gate proves canvas round-trips to a byte-identical definition
Matrix cell	Visual no-code workflow builder · `runtime/nx_wflow_ui.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_ui`; declare `nx_wflow_ui TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WA4 Inline test and edit-in-place · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfu_inline_test` · executor Organ · 1.5 u · stands on WA3
Definition of done	Run one step from the canvas against a recorded event and show its captured vars; edit a step arg in place and verify the draft without firing (the CLI verify door R12 already exists); gate proves a test run writes a ledger record tagged trial and never advances the live run
Matrix cell	U2 interaction UX (drag, inline test, edit-in-place) · `runtime/nx_wflow_ui.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_ui`; declare `nx_wflow_ui TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WA5 Onboarding, docs, accessibility · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfu_onboarding` · executor Organ · 1.5 u · stands on WA3
Definition of done	First-run guide, per-step help drawn from the organ headers, WCAG contrast on every view; measured by nx_ui_judge on the same five axes that certified R7
Matrix cell	U4 product polish (onboarding, docs, accessibility) · `runtime/nx_wflow_ui.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_ui`; declare `nx_wflow_ui TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WA6 NL-to-flow copilot · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfl_nl_to_flow` · executor Organ · 3 u · stands on R12, WA3
Definition of done	The sovereign model drafts a definition from a sentence; the draft is VERIFIED (R12 verify) before it is shown and a refused draft shows the refusal; gate proves a known prompt yields a definition that passes verify and an impossible prompt yields a refusal, never a half-flow
Matrix cell	NL-to-workflow LLM copilot · `runtime/nx_wflow_llm.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_llm`; declare `nx_wflow_llm TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WA7 RPA UI step · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfr_ui_step` · executor Organ · 3 u · stands on R8
Definition of done	A step that drives a desktop or browser surface through the estate's own input-abstraction layer (nx_input_abstract) with screenshot evidence per action; gate proves a scripted form fill lands and the evidence frames are in the ledger
Matrix cell	RPA UI-level automation · `runtime/nx_wflow_rpa.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_rpa`; declare `nx_wflow_rpa TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WA8 Record and replay · cell WATCHING · gate NO-GATE-DECLARED

Contract	`wfrec_capture` · executor Organ · 3 u · stands on WA7
Definition of done	Capture the operator's actions into a definition the engine replays (the PAD recorder category); gate proves a recorded sequence replays to the same end state via WA7 and a divergence is recorded not hidden
Matrix cell	Record-actions macro recorder (watch and replay) · `runtime/nx_wflow_record.nx` · WATCHING decl
Gate	NO-GATE-DECLARED — `knowledge/organ_gate.conf` carries no row for `nx_wflow_record`; declare `nx_wflow_record TAB <gate>` and the ship loop proves it and the verdict lands here.
Work log	no log row names this rung yet

WB1 OSWorld-Verified run of the sovereign RPA step · cell NO-MATRIX-ROW · gate NO-MATRIX-ROW

Contract	`wfr_osworld_run` · executor Organ · 4 u · stands on WA7
Definition of done	DONE-RULE PRE-DECLARED: the OSWorld-Verified 361-task set runs on an isolated Ubuntu oracle host outside the build path, driven by the WA7 RPA step through the sovereign workflow engine and graded by the benchmark's own checkers; the receipt carries n, max steps, the success rate and the host hash, and is published beside the sheet's leaders [osworldsheet26] [osworld26]; a run with fewer than 361 tasks prints PARTIAL and never the headline
Matrix cell	NO-MATRIX-ROW — no matrix row carries `wfr_osworld_run`; the ranker reads this rung as unmapped and no cell can flip for it. Add the row and the join completes on the next publish.
Gate	the gate join needs a matrix row: no organ is named for this rung
Work log	2026-09-16 land 2026-09-16 the ladder to SOTA declared as data: five dated targets (T-BIC-OSWORLD on the 2026-08 results sheet read through the sovereign XLSX reader, T-FR-TAU2 on the July-2026 grading, T-BIC-WEBARENA, T-BIC-GAIA, T-BIC-TAC on listings attested this month), 21 of 21 rungs placed (15 substrate, 1 arm, 5 contenders WB1 to WB5); nx_compare_rank reads FRESH and LADDERED; the WebArena, GAIA and TAC leaderboard numbers are stated as owed until their client-side listings are mirrored, and OSWorld-Verified's 90.19 leader is the one number read from bytes

WB2 WebArena environments through the sovereign browser · cell NO-MATRIX-ROW · gate NO-MATRIX-ROW

Contract	`wfr_webarena_run` · executor Organ · 3 u · stands on WB1
Definition of done	the self-hosted WebArena sites stand up on the oracle host and the sovereign browser drives the tasks through the WA7 step; graded by the benchmark's checkers; the receipt names the site versions [webarena23]
Matrix cell	NO-MATRIX-ROW — no matrix row carries `wfr_webarena_run`; the ranker reads this rung as unmapped and no cell can flip for it. Add the row and the join completes on the next publish.
Gate	the gate join needs a matrix row: no organ is named for this rung
Work log	no log row names this rung yet

WB3 tau2-bench domains through the connector catalog · cell NO-MATRIX-ROW · gate NO-MATRIX-ROW

Contract	`wfc_tau2_run` · executor Organ · 3 u · stands on R3
Definition of done	the tau2 domains and the March-2026 voice and knowledge extensions run against the benchmark's user simulator on the oracle host with the R3 connector catalog as the tool surface; pass-hat-k is published with k named [tau2bench26]
Matrix cell	NO-MATRIX-ROW — no matrix row carries `wfc_tau2_run`; the ranker reads this rung as unmapped and no cell can flip for it. Add the row and the join completes on the next publish.
Gate	the gate join needs a matrix row: no organ is named for this rung
Work log	no log row names this rung yet

WB4 GAIA questions through the sovereign research runner · cell NO-MATRIX-ROW · gate NO-MATRIX-ROW

Contract	`wfr_gaia_run` · executor Organ · 3 u · stands on R3
Definition of done	GAIA level 1 to 3 questions answered by the research runner with the workflow engine's tools, scored by exact match against the validation answers mirrored with licence [gaia23]
Matrix cell	NO-MATRIX-ROW — no matrix row carries `wfr_gaia_run`; the ranker reads this rung as unmapped and no cell can flip for it. Add the row and the join completes on the next publish.
Gate	the gate join needs a matrix row: no organ is named for this rung
Work log	no log row names this rung yet

WB5 TheAgentCompany tasks · cell NO-MATRIX-ROW · gate NO-MATRIX-ROW

Contract	`wfr_tac_run` · executor Organ · 4 u · stands on WB1, WB3
Definition of done	the self-hosted company (planning, chat, code hosting) stands up on the oracle host from the mirrored setup and the sovereign agent works its tasks through WA7 and R3; the receipt publishes the task completion score with partial credit the way the benchmark defines it [theagentcompany24]; OpenAgents' three-agent split (data, plugins, web) is the design reference for the tool surface [openagents23]
Matrix cell	NO-MATRIX-ROW — no matrix row carries `wfr_tac_run`; the ranker reads this rung as unmapped and no cell can flip for it. Add the row and the join completes on the next publish.
Gate	the gate join needs a matrix row: no organ is named for this rung
Work log	no log row names this rung yet

dossiers=21 cells_joined=16 no_matrix_row=5 (sum 21) gates_recorded=0 gates_unrecorded=0 no_gate_declared=16 (sum 16 of cells_joined) evgates=UNSTAMPED

Capability comparisons

Compare the field, search individual capabilities and open their source and adoption evidence. Documented presence does not establish comparative quality.

Position map — centrality and distinctiveness

The four-quadrant map the field uses for brand strategy (Dawar and Bagga, HBR June 2015), re-derived from this matrix on every publish. Centrality is the share of the category's feature mass a player covers, each feature weighted by how many hold it; distinctiveness is the average lead over each rival on the rows the player holds; breadth is the depth-weighted share of the whole matrix (the bubble); depth is how deeply the rows held are held; momentum is the day-over-day move off the spine (green rising, red falling, grey until day two); the dashed path runs first day → previous day → today. Dividers are the category means. Axes are fitted to the field of play, so read the tick numerals, not the frame. Rival marks are documented presence, so a rival's position reads the record, never its quality. The picture grades its own readability below; the table beside it is the same data for a screen reader or a second method.

Rows. 19 feature rows mapped of 34 matrix rows; 15 substrate rows (kernels, bounds and instruments the plan's rungrole marks substrate) are the estate's scaffolding, not category features, and stay off the map

Views. 2D cut: centrality, distinctiveness · 3D cube: centrality, distinctiveness, breadth · axes registered: centrality, distinctiveness, breadth, depth, momentum · a board picks its own in knowledge/compare/automation.cdmap (cut|x|y, cube|x|y|z); absent = the HBR defaults

bubble area = breadth · ring = momentum (green rising, red falling, grey until day two) · dashed = category means · axes fitted to the field of play: centrality 650–1000, distinctiveness 0–300 of 0–1000 permil (the full range put every player in one corner)
readability of the 2D cut, self-graded by the layout ruler: label overlaps 0 · labels over marks 0 · off-canvas 0 · unresolved labels 0 · mark overlaps 0 (a fact of the data: two players that close are that close) · data spread 686 permil of the plot · quadrant words unseated 0
text contrast, measured with wcag2-ratio (floors from contrast.conf), light theme: labels 16.24 (floor 4.50) · notes 16.24 (floor 4.50) · quadrant words 5.89 (floor 4.50) · tick numerals 5.89 (floor 4.50) · axis titles 5.59 (floor 4.50) · dark theme: labels 13.78 (floor 4.50) · quadrant words 5.55 (floor 4.50) · axis titles 5.97 (floor 4.50) · dark classes under their floor 0 (one figure serves both themes: a dark shortfall is a token to fix, never a class to hide) · classes refused under their floor 0 (a refused class is not drawn; the scale classes are measured, never hidden) · export: SVG PNG (receipt, rendered by the estate's own rasteriser from this page)

axes fitted: centrality 650–1000 · distinctiveness 0–300 · breadth 350–700 permil · farther bubbles are painted first, nearer ones over them · the floor shadow is each bubble's (x, y) at height 0
readability of the 3D cube, self-graded by the layout ruler: label overlaps 0 · labels over marks 0 · off-canvas 0 · unresolved labels 0 · mark overlaps 0 (a fact of the data: two players that close are that close) · data spread 366 permil of the plot

Nishi Zapier n8n Temporal Power Automate
10 panels over 5 registered axes, every pair once (the lower axis on x, the higher on y) · each panel fitted to its own field of play, first and last tick numerals shown · no labels in a small cut, the legend names the colours; the grade below reports the mark terms only (MM summed over the panels, spread averaged)
readability of the small multiples, self-graded by the layout ruler: label overlaps 0 · labels over marks 0 · off-canvas 0 · unresolved labels 0 · mark overlaps 4 (a fact of the data: two players that close are that close) · data spread 548 permil of the plot

Player	Quadrant	centrality	distinctiveness	breadth	depth	momentum	Rows held	Days on spine	First seen
Nishi	Aspirational	962	259	666	703	0 since 2026-09-30	18	12	2026-09-15
Zapier	Mainstream	901	83	456	541	0 since 2026-09-30	16	12	2026-09-15
n8n	Mainstream	975	69	473	500	0 since 2026-09-30	18	12	2026-09-15
Temporal	Unconventional	716	211	456	666	0 since 2026-09-30	13	12	2026-09-15
Power Automate	Mainstream	901	104	491	583	0 since 2026-09-30	16	12	2026-09-15

Day	Matrix rows reviewed	Players recorded
2026-09-15	34	5
2026-09-16	34	5
2026-09-17	34	5
2026-09-18	19	5
2026-09-19	19	5
2026-09-20	19	5
2026-09-21	19	5
2026-09-22	19	5
2026-09-23	19	5
2026-09-29	19	5
2026-09-30	19	5
2026-10-08	19	5

How to read this evidence. A Nishi source mark records whether the named symbol is declared in the selected source file. It does not establish working behavior or accepted delivery. A watching tag names an expected source entry; discovering it updates source coverage only. A dark tag means the file exists but the named symbol was not found: inspect the source and contract before assuming the capability is missing or renaming code. Peer codes are authored assessments; unassessed means no comparison is established, not absence. Qualification, deployment identity, runtime observations and acceptance must carry their own evidence. Adoption records the catalog's observed path through source, build, promotion, registration and invocation; libraries use importer evidence and gates use declared execution surfaces. A full catalog path does not establish quality, freshness or a successful user outcome. Partial and unknown states retain their stated next action. Census stamps: importers asof 1790637268, gate census asof 1790104131 (unix seconds; -1 = census absent).

Capability matrix — measured against source

◉ authored stronger claim; source present● present◐ partial○ absent? unassessed; excluded from relative comparison · click any capability for its evidence

Peer cell coverage: authored assessed 136, unassessed 0. Authored assessments are not verified outcomes.

Capability	Nishi	Zapier	n8n	Temporal	Power Automate
Trigger-action rules engine (rules as data) Measured: `fl_load` exists in `runtime/nx_crm_flow.nx`, verified at emit. Event-cond-action rules validated whole before load; Temporal is code-first not rules rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `SOURCE-ONLY` — PARTIAL: source exists, never compiled: /api/build it. not adopted: source exists, never compiled: /api/build it Pros Implementing source symbol found; runtime acceptance separate; ahead of TemporalCons not adopted yet: source exists, never compiled: /api/build it	●	●	●	◐	●
Idempotent exactly-once firing by event id Measured: `fl_evid` exists in `runtime/nx_crm_flow.nx`, verified at emit. Dedup ledger by event id; Temporal exactly-once execution is the reference rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `SOURCE-ONLY` — PARTIAL: source exists, never compiled: /api/build it. not adopted: source exists, never compiled: /api/build it Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Power AutomateCons behind Temporal (leads); not adopted yet: source exists, never compiled: /api/build it	●	◐	◐	◉	◐
Scheduled recurring execution (tickless + RRULE) Measured: `TLN_MAXBEATS` exists in `runtime/_hdl_build/nx_clock_tickless.nx`, verified at emit. Tickless bounded scheduler sleeps to next due job; RRULE incl YEARLY [rfc5545] Adoption: `PRODUCER-ALIAS` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	●	●	●	●
Reminder automation with once-only ledger Measured: `VIZSLA-REM-DUE` exists in `runtime/_hdl_build/nx_vizsla_remind.nx`, verified at emit. Lead-day reminders; sent-once guaranteed by an additive ledger rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `PRODUCER-ALIAS` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Temporal; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	◐	◐	○	●
Multi-channel send incl physical mail Measured: `cmd_intent` exists in `runtime/nx_send.nx`, verified at emit. Email plus print-ready postcard letter card composed in-house; rivals need Lob-class connectors rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=19 nonval=18` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Temporal, Power Automate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	◐	◐	○	◐
Bulk personalized merge to a segment Measured: `bm_personalize` exists in `runtime/nx_send_merge.nx`, verified at emit. Placeholder-validated personalization; unknown placeholder fails loud rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `SOURCE-ONLY` — PARTIAL: source exists, never compiled: /api/build it. not adopted: source exists, never compiled: /api/build it Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, TemporalCons not adopted yet: source exists, never compiled: /api/build it	●	◐	●	○	●
Document-generation action (real .docx) Measured: `dg_store_doc` exists in `runtime/_hdl_build/nx_vizsla_docgen.nx`, verified at emit. Coordination fields become a versioned Word document rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `PRODUCER-ALIAS` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Temporal; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	◐	◐	○	●
Autonomous dispatch runs registered work Measured: `sd_run_gate` exists in `runtime/_hdl_build/nx_sclass_dispatch_gate.nx`, verified at emit. Machine leases a stream, runs its gate, marks DONE only on GREEN rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `GATE:LIVE trial=-` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Temporal, Power Automate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	○	○	◐	○
Durable crash recovery (ledger replay + revive) Measured: `br_replay_resume` exists in `runtime/_hdl_build/nx_boot_revive.nx`, verified at emit. Reflog replay plus boot revive; Temporal durable execution is the reference rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `BUILT-UNPROMOTED` — PARTIAL: compiled, never promoted to the serving root: /api/promote it. not adopted: compiled, never promoted to the serving root: /api/promote it Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Power AutomateCons behind Temporal (leads); not adopted yet: compiled, never promoted to the serving root: /api/promote it	●	◐	◐	◉	◐
Run liveness heartbeats ALIVE/STALLED Measured: `hbm_scan_registry` exists in `runtime/_hdl_build/nx_heartbeat_str.nx`, verified at emit. Liveness recomputed from real beats vs the clock, not self-reported status rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=11 nonval=6` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	●	●	●	●
Ops pipeline automation with auto-rollback Measured: `ma_do_build` exists in `runtime/_hdl_build/nx_mgmt_api.nx`, verified at emit. Never-brick health-gated promote; upload-unpack-build-deploy over one API rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIVE-DAEMON` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Temporal, Power Automate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	○	◐	◐	○
Outbound event stream (SSE) Measured: `event-stream` exists in `runtime/_hdl_build/nx_mgmt_api.nx`, verified at emit. GET /api/events text/event-stream [whatwg-sse] Adoption: `LIVE-DAEMON` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	●	●	●	●
Unified multi-step durable workflow engine Measured: `wf_resume` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R1 LANDED gated GREEN: event-sourced run ledger, per-step retry, durable dedup, crash-resume; Temporal still leads depth (timers, signals, child workflows) [temporal-workflows] Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). dossier Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier; fully adopted on the estate ladderCons behind Temporal (leads)	●	◐	●	◉	●
Sovereign organ-connector catalog (any organ as a step) Measured: `wf_exec_organ` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R3 LANDED gated GREEN: exec-organ over a GREEN fail-closed catalog, proven composing the real ICS organ; n8n node ecosystem is the OSS reference rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). dossier Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons behind n8n (leads)	●	●	◉	●	●
Inbound webhook triggers Open — watching `runtime/nx_wflow_hook.nx` : `wfh_webhook_accept`, re-measured on every compare beat. Ship that symbol and this mark flips itself; the `comparewatch-` plane row flips with it. Zapier built the category [zapier-webhooks] dossierwatching wfh_webhook_accept Pros none measured yetCons behind Zapier (leads), n8n, Temporal, Power Automate; open contract; expected source declaration not established	○	◉	●	●	●
Third-party SaaS connector catalog Open — watching `runtime/nx_wflow_connect.nx` : `wfc_connector_call`, re-measured on every compare beat. Ship that symbol and this mark flips itself; the `comparewatch-` plane row flips with it. Zapier 8000-plus external apps [zapier-apps]; Power Automate 1500-plus [pa-connectors]; the sovereign-internal catalog above is ours dossierwatching wfc_connector_call Pros none measured yetCons behind Zapier (leads), n8n, Power Automate; open contract; expected source declaration not established	○	◉	●	○	●
Visual no-code workflow builder DARK — `runtime/nx_wflow_ui.nx` EXISTS but does not declare `wfu_canvas_emit`: something shipped at this path under another name, and this mark would read open forever. Repoint the contract to the real entry point (the plan rung AND this row) or rename the function; the flip follows on the next beat, and the `comparewatch-` plane row reads DARK until then. n8n node canvas is the open-source reference [n8n-docs] dossierwatching wfu_canvas_emit Pros none measured yetCons behind Zapier, n8n (leads), Power Automate; open contract; expected source declaration not established	○	●	◉	○	●
NL-to-workflow LLM copilot Open — watching `runtime/nx_wflow_llm.nx` : `wfl_nl_to_flow`, re-measured on every compare beat. Ship that symbol and this mark flips itself; the `comparewatch-` plane row flips with it. Zapier and Power Automate ship Copilot builders; n8n AI nodes partial rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them dossierwatching wfl_nl_to_flow Pros none measured yetCons behind Zapier, n8n, Power Automate; open contract; expected source declaration not established	○	●	◐	○	●
Human-in-the-loop approval steps Measured: `wf_decide` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R4 LANDED gated GREEN: approve step parks the run until an APPROVED or DENIED ledger decision lands, deny wins fail-safe; Power Automate Approvals still leads UI depth [powerautomate] Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). dossier Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier, n8n, Temporal; fully adopted on the estate ladderCons behind Power Automate (leads)	●	◐	◐	◐	◉
Process mining / run observability Measured: `wf_obs_board` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R6 LANDED gated GREEN: run board + 0-JS HTML derived by ledger replay, nothing self-reported; Temporal visibility UI still leads depth rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons behind Temporal (leads)	●	●	●	◉	●
RPA UI-level automation Open — watching `runtime/nx_wflow_rpa.nx` : `wfr_ui_step`, re-measured on every compare beat. Ship that symbol and this mark flips itself; the `comparewatch-` plane row flips with it. Power Automate Desktop leads this cohort rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them dossierwatching wfr_ui_step Pros none measured yetCons behind Power Automate (leads); open contract; expected source declaration not established	○	○	○	○	◉
Record-actions macro recorder (watch and replay) Open — watching `runtime/nx_wflow_record.nx` : `wfrec_capture`, re-measured on every compare beat. Ship that symbol and this mark flips itself; the `comparewatch-` plane row flips with it. Power Automate Desktop recorder defines this category [pad-recorder]; we have NO capture-my-actions capability anywhere dossierwatching wfrec_capture Pros none measured yetCons behind Power Automate (leads); open contract; expected source declaration not established	○	○	○	○	◉
Data passing between steps (outputs to inputs, variables) Measured: `wf_subst` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R8+R8b LANDED gated GREEN: event fields auto-bind as run vars, set-var derives, connector>var CAPTURES organ stdout, placeholder substitution into any step arg incl forked argv, last-write-wins, unknown fails loud; Zapier and n8n field-mapping UIs still lead rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). dossier Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons behind Zapier (leads), n8n (leads)	●	◉	◉	●	●
Mid-flow branching / conditional paths Measured: `wf_cond_ok` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R9 LANDED gated GREEN: per-step k=v condition vs the run variable plane, mismatch SKIPs recorded in the ledger; n8n IF/Switch canvas still leads expressiveness rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). dossier Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons behind n8n (leads)	●	●	◉	●	●
Loops / iterate over a collection Measured: `wf_for_each` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R11 LANDED gated GREEN: for-each forks a connector per comma-list item, {item} bound per iteration, WFITER records, failed item fails the step; n8n and Power Automate loop nodes still lead expressiveness rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). dossier Pros Implementing source symbol found; runtime acceptance separate; ahead of Zapier; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	●	◐	●	●	●
Error-handling branches (on-failure path) Measured: `ONFAIL` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R10 LANDED gated GREEN: exhausted step routes to a later handler step, failure stays recorded, wf_error vars set, run can RECOVER to DONE; Temporal compensation patterns still lead depth rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons behind Temporal (leads)	●	●	●	◉	●
Flow versioning / draft-test-live Measured: `wf_defs_version` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R12 LANDED gated GREEN: @version directive, runs stamp WFVER, resume records WFVERDRIFT on cross-version resume, CLI verify validates drafts without firing; Temporal replay-safe versioning still leads depth rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). dossier Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons behind Temporal (leads)	●	●	●	◉	●
Template gallery / marketplace Measured: `wf_tpl_instantiate` exists in `runtime/nx_wflow_engine.nx`, verified at emit. R13 LANDED gated GREEN: gallery index + fail-closed instantiate that validates the copied definition whole; 2 real templates ship (venue-agreement, expense-approval); Zapier's thousands-of-Zaps marketplace is still the distribution moat rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; ahead of Temporal; fully adopted on the estate ladderCons behind Zapier (leads)	●	◉	●	○	●
U1 product visual design (vs the Zapier editor) Measured: `head` exists in `runtime/nx_wflow_ui.nx`, verified at emit. R7 LANDED: nx_wflow_ui console CERTIFIED 1000permil by nx_ui_judge (DE/US/CR/AX/CO all 1000), live /compare/automation/bench; tokens + gradient + SVG + responsive + a11y; Zapier editor still leads as an interactive product rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `SOURCE-ONLY` — PARTIAL: source exists, never compiled: /api/build it. dossiernot adopted: source exists, never compiled: /api/build it Pros Implementing source symbol found; runtime acceptance separate; ahead of TemporalCons behind Zapier (leads); not adopted yet: source exists, never compiled: /api/build it	●	◉	●	○	●
U2 interaction UX (drag, inline test, edit-in-place) DARK — `runtime/nx_wflow_ui.nx` EXISTS but does not declare `wfu_inline_test`: something shipped at this path under another name, and this mark would read open forever. Repoint the contract to the real entry point (the plan rung AND this row) or rename the function; the flip follows on the next beat, and the `comparewatch-` plane row reads DARK until then. Zapier editor and n8n canvas set the interaction bar rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them dossierwatching wfu_inline_test Pros none measured yetCons behind Zapier (leads), n8n (leads), Power Automate; open contract; expected source declaration not established	○	◉	◉	○	●
U3 run information design (dashboard) Measured: `wf_obs_html` exists in `runtime/nx_wflow_engine.nx`, verified at emit. Thin: one 0-JS run board page; Temporal UI leads info design rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=30 nonval=18` — fully adopted (top of its ladder). Pros Implementing source symbol found; runtime acceptance separate; fully adopted on the estate ladderCons behind Temporal (leads)	●	●	●	◉	●
U4 product polish (onboarding, docs, accessibility) DARK — `runtime/nx_wflow_ui.nx` EXISTS but does not declare `wfu_onboarding`: something shipped at this path under another name, and this mark would read open forever. Repoint the contract to the real entry point (the plan rung AND this row) or rename the function; the flip follows on the next beat, and the `comparewatch-` plane row reads DARK until then. Mature-product polish entirely absent rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them dossierwatching wfu_onboarding Pros none measured yetCons behind Zapier (leads), n8n, Power Automate; open contract; expected source declaration not established	○	◉	●	○	●
Sovereign own-stack zero-per-task cost Source present; authored stronger claim: `ss_fold_cp` in `runtime/nx_seg_store.nx`, verified at emit. Own language compiler TLS store on your hardware; Zapier bills per task; n8n and Temporal self-host on Node and Go stacks rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=498 nonval=372` — fully adopted (top of its ladder). Pros Source symbol found; stronger capability claim is authored; ahead of Zapier, n8n, Temporal, Power Automate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	◉	○	◐	◐	○
Additive never-lose tamper-evident run history Source present; authored stronger claim: `ws_put_locked` in `runtime/_hdl_build/nx_ws_cas.nx`, verified at emit. 144 of 144 concurrent writes zero lost measured vs live git; rivals cap run-history retention rival marks uncited — the Yes, Best or Part codes on this row are an observation read with no reference mark behind them Adoption: `LIB-WIRED importers=10 nonval=5` — fully adopted (top of its ladder). Pros Source symbol found; stronger capability claim is authored; ahead of Zapier, n8n, Temporal, Power Automate; fully adopted on the estate ladderCons none on the measured axes (rival marks are documented presence, not depth)	◉	◐	◐	◐	◐

Rival-claim provenance. Every Yes, Best or Part mark in a rival column is a claim about someone else's product. A cited row carries a reference mark that resolves to a pinned mirror (the refs gate measures that); an uncited row is an observation read and is badged in its evidence until a reference lands. The badge is the receipt, never the proof: a mark proves a mirror exists, not that the mirror supports the code. The rows are counted below.

Rival claims, counted
Class	Count	Note
rows with rival marks	34	rows carrying a Yes, Best or Part in a rival column
cited	8	a reference mark resolves to a pinned mirror
uncited	26	an observation read, badged until a reference lands
rows, summed	34	RECONCILED: the parts sum to the total 34

Delivery and evidence

Inspect risks, technical debt, rendered observations, experiments and references. Read scope and limitations alongside every result.

Risk register

Risk	Likelihood x impact	Mitigation
A connector that writes to an external API is an outbound side effect the ledger must own -- a retry that re-posts is a duplicate order	likely x high	WA2 carries an idempotency key derived from run id plus step id and the gate replays it.
A copilot draft that runs without verification ships a hallucinated flow	possible x high	WA6 is verify-before-show by done-rule; the refusal path is a gate tooth, not a hope.

Release history and work log

The worked plan, on the board. Every leg appends what it measured, landed, retracted, learned and left queued, per rung, so a crash leaves the next seat a ledger here rather than a transcript to mine. Kinds: measure, land, retract, lesson, queue. Newest last.

When	Rung	Kind	Entry
2026-09-02	R1	measure	The domain's executed proof (automation.gates: nx_sclass_dispatch_gate) read RED 1/4 on 2026-09-02 while the box was saturated: its proof-gate build was REFUSED by admission, and the gate convicted the subject. Two defects, both in the gate: it hand-tallied pass/fail into a one-state gv_ctr (a dropped conjunct is invisible), and it matched the runner's PRINTED rc=4 where nx_sov_build_run exits SBR_ADMIT_REFUSED=6 for every denial (3/4/5 collide with COMPILE_FAIL/ASM_FAIL, by design). Fixed: per-tooth gv_check, gv_need SKIP on exit 6 (subject unexamined, never RED), T3 liar-kill runs regardless; rebuilt 97,515 B sha d1d0eefe, promote in flight. The referee itself was also fixed the same day (private capture path per run; exit 124 is the referee's deadline, not the gate's verdict).
2026-09-02	R1	land	nx_sclass_dispatch_gate PROMOTED 97515 B sha d1d0eefe (gv_check per tooth, gv_need SKIP on runner exit 6, T3 phantom-gate liar-kill runs regardless) and the referee nx_swcompare_evidence PROMOTED 115389 B sha 57975459 (per-run private capture path; a deadline kill before any verdict line is DEADLINE-UNMEASURED = abstain). Re-refereed 2026-09-02: dispatch gate SKIP exit=3 pass=1/1 under a REFUSED builder admission, subject unexamined; evstamp_automation grounded=26/26 gates=0/1 skipped=1 redseen=0 unproven=1 unproven_reason=every-gate-abstained. The domain no longer convicts itself on a saturated box: it abstains until the builder admits, and reads UNPROVEN not RED on /compare and in nx_deploy_ready.
2026-09-13	R1	measure	2026-09-13 native workflow evidence repair adopted and callable: nx_wflow serving and offc artifacts 110128 B SHA256 37b380449305b5fdd888faf6308a291b1903985bc512e2b9c88371ff3bfbc0c0 independently verified. Required transition journal failures stop execution; canonical gate nine failure checks plus legacy state selftest passed. Exact post-action receipt fault fixture passed 12 checks. Authored MCP schema adopted; scoped discovery reconcile first registered=1 then already=1. Actual admin API fire completed harmless variable plus retry-probe workflow; durable ledger retained, repeat event deduped unchanged, directory-ledger negative refused before action. No message or task stubs count as delivery. No public or wildcard grant change. Alpha-Beta-Production product promotion, rights authority, crash reconciliation and direct attached MCP credential validation remain unverified; directory read refusal has zero diagnostic output. Full evidence knowledge/gates/workflow-exposure-candidate-t230.json; API jobs1789309701,1789309711,1789309720. No claim of exactly-once or photoreal improvement.
2026-09-13	WA1	land	2026-09-13 WORKFLOW FAILURE DIAGNOSTICS LIVE: canonical runtime/nx_wflow.nx6167B adopted with exact5477B backup knowledge/gates/nx-wflow-before-diagnostics-t240.nx. Real candidate gate job1789309922 passed21/21 across failed fire/resume/board and successful existing-run dedup; guarded ship1789309976 installed SHA ec38e8f50da26dc52273ac0da9c5ee4a7d20a7e52fc18cb53912e7833f05cd98. Actual API negative1789309996 preserves exit1 and emits nishi-workflow-error/1 native_code-1, effects unverified, retry reconcile-first; earlier9720 silently returned zero output. No claim all validation failures structured, exactly-once effects, or automated stage acceptance. Gate provenance closure remains unbanked and ship reports independent direct-edge approximation; actual candidate executable explicitly supplied to gate. Outer organ deadline discovery: ma_do_organ_run supports form deadline_ms separately from builder --run-timeout-ms; two visual workers notified after observed12s terminations. Discovery/schema guidance needs this distinction retained.
2026-09-13	WA1	audit	2026-09-13 MCP QUEUE CANDIDATE ROOT REVIEW: actual helper gate1789310332 passes13/13 with real occupied worker, FIFO, queued/running duplicate identity, pre-dispatch cancellation, memory refusal, expiry and child recovery. No installed queue claim. Root read pending candidate27286B queue implementation and stdio candidate main; found bootstrap queue_bytes=workers*request_cap omits copied-body metadata/control/normalized-id charge. With one worker a valid near-request_cap line could newly fail despite idle executor. Returned for checked envelope derivation and real boundary regression before adoption; do not just increase constant. Actual stdio stream test still pending. Native compiler aggregate-control-copy fault discovered in first candidate remains separately retained debt, not closed by explicit-field copy. This is reviewed candidate evidence, not dependable swarm completion.
2026-09-13	WA1	measure	2026-09-13 MCP admission queue candidate qualified, NOT installed: final native helper0610 GREEN13/13 including full-budget cancellation with occupied worker; exact stdio ELF gate0638 GREEN10/10 including maximum8192-byte request under default queue, explicit128-byte configured queue refusal and zero wait startup rejection. Final candidate485471B SHA25607625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6. Root-found compatibility gap corrected by checked body plus normalized-identity plus actual metadata envelope, not raised constant. Structured queue state and correlated undispatched refusals retained. Existing allocator abort paths and compiler embedded aggregate defect0497 RED1/4 remain open; no live upstream success or attached bridge repair claimed. Review knowledge/gates/mcp-admission-queue-review-t230.json. Root deployment review required.
2026-09-13	WA1	measure	MCP queue reviewed canonical adoption0822/0823/0824 exact backups/readbacks complete. Canonical helper0870 GREEN13/13, exact canonical stdio0921 GREEN10/10; incumbent samepool recovery0878 GREEN14/14, control0894 GREEN13/13, HTTP framing0902 PASS10, actual HTTPS404 recovery/public initialize1124 GREEN7/7. Immutable knowledge/gates/nx_mcp_stdio-queue-t230.elf485471B SHA07625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6 copied0940 and hashed0956. Current running bridge untouched. Authenticated tool roundtrip still pending verified NAS credential-file path; public initialize is not authentication proof. Complete record knowledge/gates/mcp-admission-queue-review-t230.json.
2026-09-13	WA1	measure	AUTHENTICATED MCP QUEUE PROOF: immutable nx_mcp_stdio-queue-t230.elf SHA07625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6 passed native authenticated gate1470 GREEN6/6. Exact nx_fs-only grant minted via existing private-response API1344,600s lifetime, no token output. Two actual nx_fs size requests queued under default1worker then returned canonical22199B exact1 exit0/isErrorfalse; invalid-token control denied both reason2. Full positive432B and queue events independently read. Canonical source adoption already complete; record now distinguishes historical candidate limits from current open limits. Running attached bridge not switched; preserve existing credential routes and activate immutable revision through configuration owner. Review knowledge/gates/mcp-admission-queue-review-t230.json.
2026-09-13	WA1	measure	LOCAL MCP QUEUE DEPLOYMENT: immutable485471B SHA07625e4456bb0aecff9aa3438ad403d824804a3fac9bddad1ed8503263c5dce6 staged through NodeMCP exclusive copy/hash. Local authenticated gate GREEN6/6 existing configuredcap; actualcwd queuegate GREEN10/10. Exact two executable references activated via checked atomicrename in Codexconfig4809B SHA916f9d4883c8c2996f2d9b96f879615b5e4931627b64fde1d7abdb8535a14dca; all routes/caps/settings preserved; backupretained. Postactivation freshnativegate exec1960d2 GREEN6/6. Existing attached connector NOTprovenreloaded; do not claim threadtransport fixed. Narrow rule27 sovereignELF launch exception retained; no shelltransfer/Pythonwrapper. Review knowledge/gates/mcp-admission-queue-review-t230.json.
2026-09-13	WA1	measure	Freshbridge exactconfigured nishi_admin route verified: nx_mgmt GET/api/health accepted with existing admincap, terminal185B16services0down1degraded swap999. Childexited0 onEOF; no runningconnector killed. Hostreload notcallable in currenttoolinventory; normalserverRestart/freshattachment remainsrequired before claiming attachedrepair. Record knowledge/gates/mcp-admission-queue-review-t230.json.
2026-09-13	GE57	audit	2026-09-13 SEARCH OUTPUT DISCOVERY REPAIR: actual nx_shelltool grep with out=knowledge/gates/texture-owner-scan-t244.out job2162 returned158Bsummary and retained540B exact4matches. Root read full retainedresults. Existing capability omitted in knowledge/tool_schemas.conf nx_shelltool usage; corrected2183 to explain unique .out artifacts, narrowowner scope, overwritebehavior and partialcoverage caveats. Scoped existingowner reconcile2276 rows1 already1 failed0; attachedclientdescription refresh notproved. Output retention saves conversationalvolume, notscanIO (234163637B read in this observed scan). No tokenROI claim. Root reviewed MCP localconfig activation exact916f9d48 hash and freshmanagementroute worker2186; attachedprocessreload remainsunverified. No specialprovider dispatch or paidAPIusageintroduced.
2026-09-13	R3	audit	2026-09-13 CORRECTION: prior log1789312276 used GE57, which is not a rung in this automation plan; source-discovery evidence belongs to the organ connector catalog R3. Prior entry retained as history. Current attached MCP still returned admitted_workers_in_use dispatched=false during independent reads; serial retry succeeded. Fresh-config bridge qualification does not establish attached transport replacement. No concurrency recovery or estate-wide reliability completion claimed.
2026-09-13	R3	audit	2026-09-13 SWARM EXECUTION CONTRACT: retained verified workstream knowledge/gates/beach-asset-subscriptions-workstream-t251.json includes verified tool contracts, access, representative preflight, existing owner repair scope, input/output/acceptance and review evidence. Contract recorded; automated enforcement not yet delivered. Actual Claude control probes established enforced available-model list prevents alternate model selection without inference. Existing adapter candidate now waits for same-child initialization and fresh billing/account evidence before user message; root reviewed full worker and requested lifecycle/protocol negative tests. No provider task launched or tokenROI claimed. Public source observations retained asset-source-observations-t253.json; first task is supplied-evidence analysis, not claimed live browsing. Shared fetch classifier native regression moved1/3 to3/3 job1789314037; broader and live validation pending. Beach inspection integration proceeds separately in existing shared owner.
2026-09-13	R3	review	2026-09-13 Root reviewed full Claude preflight candidate4a711b52c3c4a9fb7fc359a86d59af4368d47436e468d6a023ae6a555d0f36e5, exact diff and qualification21/21 protocol plus4/4 receipts. Approved first public supplied-evidence unit through existing host facade over MCP, actualaster/Fable5.1, toolsdisabled, no paidfallback. No claim canonicalWindowsadoption/native registration or automatedexecutionBriefvalidation. Actual provider result stillpending. Correction to prior livefetch failure interpretation: witness caller1MiB outputcapacity exhausted, underlyingdeflateoverflow; generic-33 mapping obscured it. Not evidence malformedgzip or productiondefaultregression. Worker retaining measuredcompressedpayload and repairingclassification without arbitraryhigherlimit. Root review files claude-preflight-qualification-t251.json and fetch-status-repaired-review-t252.json.
2026-09-13	R3	requirement	OPERATOR investment clarification: prioritize SOTA quality and timely delivery; optimize by growing Nishi's repeatable execution and evidence capabilities, not reducing quality, detail or required review. Evaluate combined accepted user outcome and reduced repeat coordination. Source presence or a green isolated test alone does not establish reusable delivery. Measure real workflow invocation, acceptance evidence, root supervision effort, provider usage and runtime cost separately; unknowns stay unknown. Current focus remains Beach delivery and supervised ingestion/swarm dispatch; no new parallel domain opened by this clarification.
2026-09-13	R3	delivery	2026-09-13 Root native release-manifest published Beach Alpha inspection-distance page d83d169c1d223f12be6ab1dfdeb751872427e42ceea941d8fbbd73410c6b8647; actual Chrome served hash and zoom/Fit/gameplay restoration verified, errors0. Evidence knowledge/gates/inspection-distance-served-root-t260.json. This is diagnostic inspection delivery, not character beauty or Beta acceptance. Shared standards required: failure categories remain distinct through consumers; source/build/configured/running/served identities must not be conflated; release preparation derives missing registration from actual dependency registry and preserves hash guards. Earlier manifest existing-only registration refusal retained715caafd; exact reviewed registration via existing release workflow succeeded352d2792. Existing owners tasked to make preparation/reload verification reusable; automatic completion not claimed. No provider trials resumed.
2026-09-13	R3	measure	2026-09-13 Shared fetch capacity repair adopted through nx_organ_ship5704 into nx_https_get_cli serving/offc SHA256 d858f100637f21244f444b1665566aa7733807cca08e965333b0bc91343e0be1 577560B; independent live hash5744 matches. Reusable rule: distinguish caller capacity exhaustion (HF_DEC_CAPACITY -34), malformed compression (-33), framing/dependency errors, and incomplete captured evidence; never treat decoded-too-large as raw-success. Native retained RigModels gzip expands306799 to2439009B with CRC and byte equality; boundary gate5219 passes13checks; live default5228 returns200 and fulldecodedbody. Actual attached MCP5743 decodedHTML but output contains explicit CAPTURE-TRUNCATED marker at1048576B; classify incomplete, not complete ingestion. Owner runtime/nx_tools_api.nx1663..1690 caps async capture and marker lacks structured completeness. No cap increased. Source fixes do not prove all deployed shared consumers rebuilt. Existing decoder MAX_BLOCKS exitwithoutfinal still falsely reportsOK: queued against runtime/nx_deflate.nx, not repaired here. Canonical webscraping/dataio Compare owner unverified; integration evidence retained here. Full review buildroot/knowledge/gates/fetch-capacity-review-t253.json.
2026-09-13	R3	measure	2026-09-13 Character source-descriptor/rig-preservation operation nx_gltf_render_prepare registered5826 through existing API alias to existing nx_gltf_render_prepare_candidate_t230.elf; no duplicate implementation. Guarded promotion5795 self-gated exactfreshartifact with retainedCatwalk GLB, serving/offc SHA2565ea8ce60601bc19fbc5a1aa727013bc21248bc37795ec865f57c9bea30c18c63 143320B hashes5827/5851. Actual API invocation5844 returned34024B descriptor c8f1397ce722308ec280b54ea77b4ca617aa279ab5bf48a81b22306869eef3e0 reused1. Discoveryreconcile5845 registered1 failed0; authoredschema title includes arguments and scope. Attached-client newtool exposure unverified. Initialship5737 refused because operator omitted requiredgatearguments; retained as invocation failure, not subjectdefect. This preserves existing source rig; not autorigging, appearanceimprovement, Beachintegration, or fullingestionworkflow. t247 boundaryreview retained; nestedallocation/durability limits remain.
2026-09-13	R3	verify	Public discovery now verified at https://nishifamily.com/api/tools job1789315927; retained784543B output contains exact nx_gltf_render_prepare entry with full argument/scope description near offset780043 and JSON end. Actual API invocation5844 alreadyverified. Attached direct-tool handle refresh remains separate. Full adoption record buildroot/knowledge/gates/fetch-prepare-adoption-t253.json. No Compare public regeneration claimed.
2026-09-13	R3	verified	2026-09-13 Root independently invoked registered nx_gltf_render_prepare via native mgmt API against retained source GLB, mesh0 skin0, exact existing descriptor: job1789316130 exit0 reused1 bytes34024 SHA c8f1397ce722308ec280b54ea77b4ca617aa279ab5bf48a81b22306869eef3e0. Full discovery JSON parsed complete1946tools with exactly1 operation row; source-descriptor/rig scope explicit, no autorigging or Beach-placement claim. Poll claim metadata before requesting output; full registry output is unnecessary context expense, project exact known record from retained evidence. Reusable API delivered; attached tool refresh, integration and consumer acceptance remain separate.
2026-09-13	R3	measure	2026-09-13 Existing native nearest-surface asset gate now adds six independently provable off-surface support-plane probes per asset, exact nearest coordinates and squared distances. Separate Paladin5015, DarkKnight5048, Toon5055, Witch5063 each45/45; resident5082 40/40. Expanded analytic gate5073 30/30 includes tilted plane, reverse-winding edge and duplicate-vertex degeneracy. Retained complete jobs and knowledge/gates/bvh-nearest-review-t284.json; asset ELF64613B93285f82bee0ad8bb9bdc4be1181bf692809b02cfb0a64372b106c6fb8fe807e. This improves reusable native evaluator independence beyond shared-kernel exhaustive agreement; no signed contact, posed motion, hair generalization, FPS or visible Alpha improvement claimed. No public Compare regeneration. Graphics domain owner not resolved in this bounded task; reusable invocation evidence retained here.
2026-09-13	R3	blocked	2026-09-13 Writer artifact load repair native21/21 on actual1415858B BeachHTML plus prior bound-request26/26 and shared47/47. Exact size-derived allocation replaces accidental1MiB display-limit input ceiling; source36676B3a7da1ac8e8bd39c5992fa8f01d6db4cbc9f19dfca07be84509b2c80688ca06b. NOT deployed: guardedship7154 caught stale gate assertions, corrected native7275 18/18; retry7348 refused17 printable losses revealing livewriter predates bound-fingerprint source. Native current-ledger7496 confirms malformed writer row32828..33102 SHA7463b3df779b8578fde7b342a26404ed07a373a53c301faf94293aff14e799c6, unknown original keys/pairing. New strict reader would block unrelated fresh keys; regression7533 explicitly proves this. Existing explicit-epoch-design-t219 recovered for safe additive namespace plan, not silent history deletion. Complete current recovery review knowledge/gates/writer-live-ledger-recovery-review-t293.json; no live journal mutation/allowance/ship or public Compare regeneration. Reusable lesson: source qualification plus isolated tests must be joined to actual retained state compatibility before deployment.
2026-09-16	WB1	land	2026-09-16 the ladder to SOTA declared as data: five dated targets (T-BIC-OSWORLD on the 2026-08 results sheet read through the sovereign XLSX reader, T-FR-TAU2 on the July-2026 grading, T-BIC-WEBARENA, T-BIC-GAIA, T-BIC-TAC on listings attested this month), 21 of 21 rungs placed (15 substrate, 1 arm, 5 contenders WB1 to WB5); nx_compare_rank reads FRESH and LADDERED; the WebArena, GAIA and TAC leaderboard numbers are stated as owed until their client-side listings are mirrored, and OSWorld-Verified's 90.19 leader is the one number read from bytes
2026-09-19	R1	audit	2026-09-19 dynamic identity repair pending: native gate1789860000 proved distinct long event IDs alias under canonical wf_evid truncation; 27/27 containment/result tests do not close full identity. nx_wflow_engine wf_fire, wf_resume, wf_decision, wf_emit and observer formatters retain fixed identity buffers. Reuse nx_wflow_text_owned measured allocation; reviewed versioned lossless new RID plus evidence-qualified legacy lookup must preserve existing journal bytes and refuse uncertain historical binding. No new engine adoption yet.
2026-09-19	R3	audit	2026-09-19 reusable part-to-stage delivery: knowledge/wflow/component-release-preparation.flows and .steps already bind product/component/stage and exact manifest path/SHA; nx_site_publish_lib rm2_preflight/execute/publish accepts generic consumer source/path/hash/bytes/expected_live and immutable dependency collection. Actual Alpha experimental publication8283 reconciled ledger and served hash; DONE is not visual acceptance. Remaining gap is candidate/component plus assembly artifact and manifest identity bound to stage-specific prerequisite evidence and approval; scope labels alone do not enforce lineage. v2 requires at least one dependency. No claim all component types or Beta/Production delivery qualified.
2026-09-19	R4	audit	2026-09-19 sequenced repair after identity: reuse existing wf_decision/wf_decide parked approval with exact candidate/component/assembly/manifest/stage evidence; reject stale approvals. Experimental Alpha may expose declared failed evidence; Beta integration/rights and Production acceptance require distinct policy. No new service and no uniform all-green rule.
2026-09-19	R6	audit	2026-09-19 command outcome boundary: native nx_body_proc unknown mode job1789860262 returned tools envelope isError false and empty text but meta exit_code7; transport success is not command success. Result adapter must preserve nonzero exit, uncertainty, full evidence references and action. Prior Claude attribution for these owners remains unverified.
2026-09-20	R1	verified	2026-09-19 dynamic workflow identity/result closure now canonical and serving: promotion1789864076 live/offc197664B SHA4c570b50b807d6bbb96f05cba273139443b0eb9c93d78ae7b39853d352ecf012, identity44/44 and scoped ownership6/6. Original event-result1789864130 read-only response1357B, historical ledger unchanged6602B; command_execution_reaped0 and served_acceptance unverified explicitly retained. Full evidence buildroot/knowledge/gates/alpha-release-workflow-delta-t343.json. This supersedes candidate status above, not history.
2026-09-20	R1	debt	2026-09-19 native boolean boundary evidence: probe1789863739 evaluates RHS once for both true OR probe and false AND probe. Wrapper1789863616 crashed after assuming short-circuit argv guard; explicit nested argc branches qualify44/44 in1789863672. No compiler bug asserted: shared review/lint must flag guard-and-dereference assumptions under measured language semantics. Evidence buildroot/knowledge/gates/workflow-eager-boolean-evidence-t344.log and workflow-argv-boundary-failure-t344.log. No estate-wide lint coverage claimed.
2026-09-20	R1	debt	2026-09-19 repeated allocator-contract defect: shipping behavior candidate t371 initially used aborting sys_mmap followed by null checks for argv/path scratches (source delta cb7d5094, compile7693); those structured refusal branches were unreachable on allocation failure. Prior workflow ownership review independently established sys_mmap_try/direct-unmap contract. Corrected current owner nx_organ_ship_review_candidate_t371 uses one checked argc/path extent and a sys_mmap_try cleanup wrapper through every inner return; compile7807 actual usage exit2 verified, allocation/workload gate still pending. Existing shared source-review/lint capability must detect aborting allocation followed by recovery checks and distinguish arena from direct ownership; no estate-wide detector coverage or leak-freedom claim. Exact repair/evidence: buildroot/knowledge/gates/shipping-behavior-review-t371.json.
2026-09-20	R1	debt	2026-09-19 candidate-only authority conflation caught before adoption: t371 renamed loss_approved to live_recheck_required, then inherited use also sent allow_capability_loss=yes for every successful no-loss or behavior-reviewed adoption. Root source review rejected this; new exact_loss_approved output is set only by successful os_loss_exact, separately from live identity recheck. Existing shared boundary review/lint must flag verification/observation states reused as authorization to weaken another gate. No deployed authority expansion claimed. Actual promotion-body countertests and full correction retained in buildroot/knowledge/gates/shipping-behavior-review-t371.json.
2026-09-29	R1	land	2026-09-29 ENABLEMENT: native owned PM transport installed and discoverable as nx_pm_owned_native. Exact serving/offc ELF 557321 bytes SHA256 b89678b182fee20e81160f23357e38cac294c5b20eed0a852e3640391e550996; native NTH1 host policy 594 bytes SHA256 7f4d75585bf99bb9ea2ae3523033228c4f8cf302f3593694c423954226a692cb. Existing native approval/input/version/claim owners enforce current signed operator identity, scoped capability, registry, RACI, exact version approval and replay. Private qualification job1790661374 passed43/43 with real native worker, swaps, role/approval withdrawal and scoped fixture cleanup. Install1865, register1938 and one-row discovery2070 completed; actual served invalid-intent probe1982 returned structured400/exit2 with execution_attempted=0, an expected negative test. This is installed transport, not an adopted edit-build workflow, autonomous repair, net resource savings or Beta/production acceptance. Positive current-host credential invocation and end-to-end owner workflow adoption remain open; Claude hook candidate remains uninstalled. Retained worklog buildroot/runtime/clmem/project-owned-alpha-native-architecture-2026-09-28.md. Browser connection discovery returned no available browser at2026-09-29T06:08Z; visual release review remains pending, not waived.
2026-09-29	R3	land	2026-09-29 REPAIR: Claude result hook installed with qualified ELF120858 bytes SHA256058238dd6a35e5b7ffb4ae60cbda48852d99faeab94f5999317ca075868463de. A nonzero process exit no longer asserts command failure. Native qualification157/157 passed on NAS and local host; unchanged configured Claude launcher passed11 cases/112 assertions and both exact reported jobs1790654753/1790654870 with24 assertions. Those searches returned PRESENT with590 and18 matches, complete coverage. Existing estate gate also replayed those exact client projections21/21 in job1790663486. Canonical source18789e09 adopted through native immutable install intents; canonical build3119 reused byte-identical qualified artifact. Prior source/executable and explicit reverse intents retained; install replay reconciled without another replacement. Command outcome remains UNCLASSIFIED until its contract is applied; no automatic repair/retry, active-provider-session observation or estate-wide enforcement claim. Existing debt1790624360 remains open for owned workflow/typed outcome closure. Evidence knowledge/gates/truncwarn-adoption-20260929.md and existing architecture worklog. Release visual review remains open because no browser connection was available.
2026-09-29	R1	land	2026-09-29 ENABLEMENT: native PM credential handoff installed and callable as nx_pm_native_handoff, ELF255547 bytes SHA2566fd8868ddf94135bb4a0ae1aa02f81458ed0c60a7dfbbbc92ea9b40fff2acd09. Private25/25 gate1790665106 verified real signed identity, exact workflow capability, current roles/revocations, complete input, safe replay, private output and cleanup28 files/324595 bytes/remaining0. Actual host5163 retained351-byte native NAC1 without executing a workflow; installed API5394 and direct MCP5459 reconciled the same artifact without replacement or secret output. JSON is used only for management/MCP interoperability; native output and existing NTH1 authority remain. Registered5294; scoped discovery5395 rows1/failed0. Existing workflow binding was not migrated. The complete owned edit-build-qualify-apply-verify loop, typed command semantics and automatic credential retirement remain open; no autonomous repair, net savings, Beta or frontier claim. Detailed retained report knowledge/gates/native-handoff-adoption-20260929.md; architecture worklog project-owned-alpha-native-architecture-2026-09-28. Browser recheck07:05Z returned none; visual release review remains open.
2026-09-29	R1	land	2026-09-29 ENABLEMENT: native PM staging and approved selection installed as ELF0a22b25139b0f20705769c6419bf3022f5dec678a90dda80e3179491d5e020c4 (585529 bytes), exact build/root/_offc identity verified. STAGE retains native candidates without approval or selection; SELECT requires current exact approval and compare-and-swap; admitted requests keep their implementation and rollback replay cannot reapply an old transition. Private35/35 plus transport43/43 passed. Actual MCP8145 verified installed input refusal after owner-led renewal of an expired entry credential; no automatic healing claim. Real native artifact verifier separately passed25/25 with signed stage/approve/select/admit/invoke and retained match/mismatch/unavailable outcomes; current-host workflow adoption is still OPEN and verifier is not installed. Qualification cleanup330 private files/3076771 bytes/remaining0; no net estate savings claim. Report knowledge/gates/native-stage-select-adoption-20260929.md. Native authoring API, canonical input budget admission, full owned repair workflow, credential retirement and browser visual review remain open; browser inventory empty07:50Z. Alpha; no Beta/frontier/autonomous repair claim.
2026-09-29	R1	land	2026-09-29 ENABLEMENT: native authoring installed in nx_pm_owned_define (792636 bytes SHA2561359391b8acf8aeb49769b3522900798dcbe6f70d8ae5b29f6d7326b1c13d92c), qualified29/29. Existing Nishi owners now author native versions, six-action intents and authority-preserving host budget proposals. Current-host estate-native-control-verification-v1 completed signed stage/approve/select/admit/invoke: three installed artifacts matched; exact replay reused receipt af6f17bc without execution. Native policy read budget262144 applied with exact forward and retained reverse intent. Existing discovery reconciler gained scoped refresh-v1, qualified20/20 and installed a01e1bff; live refresh changed one description and repeat held generation4142, with registration/index unchanged. Actual served MCP contracts verified09:01Z. Original private test failures retained and corrected; qualification fixtures removed with verified absence. Detailed report knowledge/gates/native-authoring-adoption-20260929.md and retained architecture worklog. Full owned repair/build/qualify/install/verify, general native receipt inspection, credential retirement, legacy bulk discovery limits and visual release review remain OPEN. Browser inventory empty08:57Z; served bytes do not establish visual acceptance. Alpha; no autonomous recovery, Beta/frontier or net savings claim.
2026-09-30	R1	land	2026-09-30 ENABLEMENT: native PM retained-request inspection is installed and exercised through the existing management API. INSPECT-only adapter702088B SHA826e3340031289b4f9750bb93827afbb54a9170f01352593f82d7a887be540ef and shared transport702104B SHAc5034f4821207c7e16c706a26a62dcb6c5e7d5bcd9ed718043152b051a82213b have verified root/offc equality. Current private inspection gate54/54 rejects all six effect actions at the read-only entry even with execution credentials; separate read scope, original actor, budgets, version retention, failure/expected-nonzero distinction and no-rerun behavior are verified. Installed author, scope registration, private grant mint and handoff drove actual inspection1790782421 and repeat1790782664: exact historical receipt af6f17bc retained, claim/generation unchanged, worker execution0, read1916005B/21 files. This is historical Sept29 ARTIFACTS_MATCH evidence, not a fresh current-artifact check. Native NTH2 explicit budgets installed with reverse intent; handoff/definition updates installed and qualified. New definition-gate cleanup45/45 includes foreign-file refusal and repeat-safe own-fixture removal. Discovery contracts and index membership verified; provider caches not claimed refreshed. Detailed report knowledge/gates/native-receipt-inspection-scope-20260929.md; retained architecture worklog project-owned-alpha-native-architecture-2026-09-28. Open: complete owned edit-build-qualify-apply-verify/close workflow, automatic credential retirement, long-lived allocation accounting, broader consumer adoption, original incident closure and comparative resource return. Browser verification through registered NAS capture1790782866 reported CD-BROWSER-ABSENT: configured candidates target Windows/WSL, no resolved browser on executing host. Visual/interaction review remains OPEN; served bytes do not close it. Alpha; no self-healing, estate-wide SOTA, Beta or net savings claim.
2026-09-30	R1	land	2026-09-30 ENABLEMENT: the existing Nishi shipping owner now authors and consumes native NGP1 per-gate inputs. Installed nx_organ_ship329176B SHA31f1ea03162326481b19e78db3bda563cb3cdba70941dbb65efeccbd03143295 has independently matched serving/runner copies. New gate62/62 and existing71/71 plus151/151 passed. Installed author replay5249 retained the same1368B native recipe172a5440 with no duplicate; real consumer build1790785287 used13 inspection inputs and8 definition inputs separately, passed54 plus41+4 checks, and promoted nx_pm_owned_define884592B SHAb44f48328efdc001bc7cbb993bec96083c0223bb5916cf49c60efebef3082b5b with independent serving/runner equality. Exact input/config identities, required-gate coverage, literal arguments and per-execution ownership are enforced; malformed, conflicting and over-budget records refuse. Source-closure banking6066 produced byte-identical qualified native gate79928f66; current closure independently checked6384. The definition gate now emits its canonical verdict after all cleanup, qualified47/47 in6334 and accepted by the original build API6383 as375600B SHAa803cfd5. Its100 isolated test files/21686B were removed and directory absence independently confirmed. These gate artifacts are qualification-only; no separate gate installation claimed. Discovery row refreshed4158; provider caches unverified. Detailed retained worklog buildroot/runtime/clmem/project-owned-alpha-native-architecture-2026-09-28.md names original evidence, rollback copies, superseded candidates and fixture debt. OPEN: automatic permanent gate workloads with legacy-call compatibility, old shipping-gate fixture disposition, full PM-owned repair/build/verify/recover/close, credential retirement, original incident closure and comparative resource return. Whole-conf recipe invalidation remains strict. Browser execution-host/config integration and served visual/interaction review remain OPEN. This is one installed and exercised build-composition capability in alpha, not autonomous recovery, estate-wide SOTA, Beta or measured net savings.
2026-10-01	R1	measure	BARFRESH 2026-10-01 (seat residue-fleet, T0 1790856940): before STALE with 6 of 6 bars UNATTESTED on the October clock; after 5 ATTESTED and 1 UNATTESTED, verdict still STALE. Attested from bytes re-fetched over the sovereign fetcher and read this month, all five byte-identical to their 16 Sep mirrors: SB-OSWORLD (results sheet re-read through nx_xlsx, 1077 rows, largest Date serial 46235 which is 2026-08-01), SB-TAU2 (README, newest dated item July 2026), SB-WEBARENA (README, newest 12/20/2024), SB-TAC (README, newest tag 20241208) and SB-OPENAGENTS (README, newest news 2023 Oct 26). Left for the arXiv attest runner and NOT fetched by this seat: SB-GAIA (gaia23, an arxiv.org abstract page). The page needs a regen to show the October barscans.

On these two registers. Rows are declared in the domain's plan file and carry the debt id, which is the join key back to the sovereign debt plane — that plane, not this page, is the authority on state. Reconciling them automatically (the regen reading the plane and refreshing these rows) is a named, owed rung; until it lands, treat an id here as a pointer to look up, not a status to trust.

Honest verdict. The coverage above records source-symbol presence. Stronger comparisons are authored claims, not measured runtime superiority or accepted outcomes. A watch contract names an expected source symbol; finding it changes source discovery only. Qualification, deployment and runtime acceptance require their own evidence.

Person · product · place — not yet measured for this domain

Every compare carries this layer. Declare knowledge/compare/automation.ppp (rows surface|nishi or c1..c4|label|url|connect naming OUR live surface and each rival's front door), run nx_ppp_probe domain automation, and this section fills itself on the next beat: the same ruler on both sides — privacy and CX (third-party hosts, tracker classes, cookies, security headers), design and longevity (design hygiene, computed WCAG contrast, render-blocking resources, unsized media, script weight, theme and motion queries), findability (landmarks, skip link, on-site search, breadcrumb, headings, internal links).

The field — discovered, not chosen

Rows written by nx_field_discover from automation.seeds: the industry's own lists (Wikipedia wikitext, GitHub topics, awesome lists) read mechanically, every candidate counted across seeds. The matrix columns above are a SEAT'S pick; this band is the population they were picked from, and the stats line measures one against the other. A rival here is a lead, never a verdict — it earns a column when its capabilities are read and pinned.

Discovery receipt. One row per counter the run wrote, in the receipt's own order; the rival rows follow in their own table.

Field discovery, counted
Class	Count	Note
seeds	4
scanned	4
fetched	4
reused	0
failed	0
named	4
candidates	4
mentions	4
capped	0

Rank	Rival	Seeds	Mentions	First seed	Kind	Link
1	Zapier	1	1	col0-zapier	named	https://zapier.com/apps
2	n8n	1	1	col1-n8n	named	https://docs.n8n.io/
3	Temporal	1	1	col2-temporal	named	https://docs.temporal.io/workflows
4	Power Automate	1	1	col3-power	named	https://learn.microsoft.com/en-us/power-automate/

Exclusive and shared capabilities

One capability is graded on one board; every other board references it. The fleet duplicate census (nx_rivalcensus dupes, on the fleetdupes clock row) reads every board's matrix and names each capability graded as a row on two or more boards, with a home derived from evidence: rung when exactly one board's plan contracts the symbol (that board is starred), gates when no plan contracts it and exactly one board's executed gate map runs the organ's gate (that board is plussed), ambiguous when several boards carry the deciding evidence (a decision for their seats), and none when no evidence decides it, with the reason beside it: unmapped (no board of the duplicate declares a gates map; declare the maps) or unproven (a map exists and runs none of the organ's gates; gate the organ on its proving board, or declare it a shared library by policy). A home elsewhere means this board should carry an @xrow reference to the home board's row rather than a graded row of its own; a home here means the other boards should. The table is this board's slice of one estate-wide census: the shape is shared, the rows are the board's.

This board grades 34 capabilities: 28 exclusive to it and 6 shared with other boards. Shared rows on this page: 6 = home here 0 + home elsewhere 0 + ambiguous 0 + undecided 6 + refused 0 (partition 6); undecided = unmapped 0 + unproven 6 + unevidenced 0; homonyms 0; since the previous census: new 0, resolved 0. Census FRESH age_s=6313.

Capability	Organ	Boards grading it	Home	Rule	Evidence	Standing on this board
ss_fold_cp	`nx_seg_store.nx`	search, webscraping, deepresearch, automation	—	none	unproven	undecided
VIZSLA-REM-DUE	`nx_vizsla_remind.nx`	automation, projectmgmt	—	none	unproven	undecided
dg_store_doc	`nx_vizsla_docgen.nx`	automation, projectmgmt	—	none	unproven	undecided
hbm_scan_registry	`nx_heartbeat_str.nx`	automation, projectmgmt	—	none	unproven	undecided
event-stream	`nx_mgmt_api.nx`	automation, deploy, containers	—	none	unproven	undecided
ws_put_locked	`nx_ws_cas.nx`	automation, projectmgmt	—	none	unproven	undecided

Gaps from the record — what the estate does that no board carries

The record census (nx_goalmap record) reads the invoked-tool population and every plan queue row and files each organ or directive that NO matrix, plan or gates row names. A row here is a callout the boards missed: adjudicate it onto a board or declare it infrastructure. Census state STALE (age 1395418 s), sources read 4 of 7 declared — a BLIND census is a FLOOR: unread sources can only add rows.

kind	name	board	source	evidence

rows shown 0|this board's directives 0|estate-wide un-boarded organs 452 (listed in full on /compare/ecosystem)|census rows 639|malformed 0 (counted, never rendered)

References

Beyond a link list. Every reference below resolves twice — the publisher's copy and, where banked, the estate's own non-rottable library mirror with a content pin — and carries its evidence class plus the exact claim on this page it grounds. Keyed marks like [key] in the matrix notes jump here. A dash means honestly absent, never assumed.

[zapier-apps] Zapier. Integrations, App and Software Automation -- the Zapier app directory (9,000+ apps at capture). Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_automation_zapier-apps.html · pin h533ee69615af3b8af5487dc2269af3106c9e0975a6aa1519473fc3389ee0b7a1 · accessed 2026-08-18 · vendor-docGrounds: The Third-party SaaS connector catalog row (Zapier 8000-plus external apps, Best -- the directory page states 9,000+ at capture, so the row's figure is a floor) and the Template gallery / marketplace row (Zapier's thousands-of-Zaps marketplace is the distribution moat).
[zapier-webhooks] Zapier. Webhooks by Zapier integrations -- catch-hook triggers and custom requests. Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_automation_zapier-webhooks.html · pin h21a07837b052cd0c0c3d6baefce12bd901aa858f64169d3de56800f694e365ba · accessed 2026-08-18 · vendor-docGrounds: The Inbound webhook triggers row (Zapier built the category; nx_wflow_hook is absent) and the U1/U2 rows naming the Zapier editor as the interaction bar.
[n8n-docs] n8n. n8n Docs -- workflow automation platform: node canvas, core nodes (IF, Switch, Loop Over Items), integrations, self-hosting. Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_automation_n8n-docs.html · pin h81d9268f40f20e0f65cb985e03614f18d4f7ce1be797e0f3b0e1aa712dff0475 · accessed 2026-08-18 · vendor-docGrounds: The n8n column: Visual no-code workflow builder Best (n8n node canvas is the open-source reference), Sovereign organ-connector catalog (n8n node ecosystem is the OSS reference), Mid-flow branching (IF/Switch canvas) and Loops (loop nodes) where n8n reads Best or Yes.
[temporal-workflows] Temporal Technologies. Temporal Workflow -- workflow definitions, determinism and replay, timers, signals, child workflows, versioning (Temporal documentation). Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_automation_temporal-workflows.html · pin h0e0c2bf7fdea275f32e7f24589a2eef4e4de973e8939c7adfb2f93135d66588d · accessed 2026-08-18 · vendor-docGrounds: The Temporal column: Idempotent exactly-once firing (Temporal exactly-once execution is the reference), Durable crash recovery (Temporal durable execution is the reference), Unified multi-step durable workflow engine (Temporal still leads depth: timers, signals, child workflows), Flow versioning (replay-safe versioning) and Error-handling branches (compensation patterns).
[powerautomate] Microsoft. Power Automate documentation -- cloud flows, desktop flows, approvals, connectors, Copilot (Microsoft Learn). Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_automation_powerautomate.html · pin h04dfd34f272ff06ae24ab3e4d1556b130448a1120c9b2f607f131e6d24922c18 · accessed 2026-08-18 · vendor-docGrounds: The Power Automate column: Human-in-the-loop approval steps (Power Automate Approvals still leads UI depth), NL-to-workflow LLM copilot (Copilot builders) and RPA UI-level automation Best (Power Automate Desktop leads this cohort).
[pad-recorder] Microsoft. Record desktop flows -- Power Automate Desktop recorder (Microsoft Learn). Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_automation_pad-recorder.html · pin h47e8fc0aa0dc8fb36618777b2a8432af13256590e890767d008a4292bf0d02c2 · accessed 2026-08-18 · vendor-docGrounds: The Record-actions macro recorder (watch and replay) row: Power Automate Desktop's recorder defines the category (Best); the estate has NO capture-my-actions capability anywhere, which this page makes concrete.
[pa-connectors] Microsoft. Connector reference for Power Automate -- the full list of standard and premium connectors (Microsoft Learn). Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_automation_pa-connectors.html · pin hec0763e550f617ee3b900fc32f54ceab7250aee17dcdc1e4d332dfa2d97b7a54 · accessed 2026-08-18 · vendor-docGrounds: The Third-party SaaS connector catalog row's Power Automate figure (1500-plus): this reference page is the enumerated connector list that number is read from, beside the Zapier directory.
[rfc5545] Desruisseaux, B. (ed.). Internet Calendaring and Scheduling Core Object Specification (iCalendar). IETF RFC 5545, September 2009. publisher · read in our library knowledge/fetched/cmp_automation_rfc5545.html · pin h8787267ca21cfbe6ae7290a0548336fb06506f90e06ba8cdd878e75b6dea6053 · accessed 2026-08-18 · published-standardGrounds: The Scheduled recurring execution (tickless + RRULE) row: RRULE including YEARLY is the iCalendar recurrence-rule grammar of section 3.3.10, so the tickless scheduler's schedule language is a published standard rather than an in-house format.
[whatwg-sse] WHATWG. HTML Living Standard, section 9.2 Server-sent events -- the text/event-stream format and EventSource. Accessed August 2026. publisher · read in our library knowledge/fetched/cmp_probe_whatwg_sse.html · pin h56e4bbeb7c54f2f55a01cdfd9cbc68ec25bb03849caa18814690b3049573761b · accessed 2026-08-18 · published-standardGrounds: The Outbound event stream (SSE) row: GET /api/events text/event-stream is this section's wire format, the same one every column's event feed speaks. Mirrored 2026-08-18 once the sovereign TLS-1.2 leg was root-fixed (debt 1787075987: the hello offered sha384/sha512 the SKE verifier could not check).
[osworld26] OSWorld project page (xlang-ai), fetched and read 16 Sep 2026: the computer-use benchmark over real desktop applications; OSWorld-Verified announced 2025-07-28 with fixed tasks and updated results, OSWorld 2.0 announced 2026-06-26; the leaderboard is rendered client-side from a results sheet the page loads. publisher · read in our library knowledge/fetched/cmp_agents_osworld.html · pin hbf37e00ce23e8c5472d485a6eb3b291ffa2de220b44040402516e41cbc48c516 · accessed 2026-09-16 · vendor-docGrounds: RPA UI step
[osworldsheet26] OSWorld-Verified results sheet (static/data/osworld_verified_results.xlsx), fetched 16 Sep 2026 and read through the estate's own XLSX reader (nx_xlsx): 1077 rows of model, institution, approach type, max steps, date and success rate over 361 tasks; the top rows at 100 steps read Intelligence-Indeed Agent 90.19 (2026-07-25), claude-fable-5 85.96 (2026-08-01), Pointer Agent with Opus 4.7 83.64 (2026-05-21), claude-opus-5 83.39 (2026-08-01), Coasty CUA v1 82.81 (2026-07-01), Holo3-35B-A3B 82.56 (2026-04-20); newest entry 2026-08-01. publisher · read in our library knowledge/fetched/cmp_agents_osworld_verified_results.xlsx · pin hcf6b4b67eed566ddcd5a8b7ad2d89013157978dad0eea76ea12d2377471efc09 · accessed 2026-09-16 · datasetGrounds: RPA UI step
[openagents23] OpenAgents README (xlang-ai, arXiv 2310.10634, Apache-2.0), fetched and read 16 Sep 2026: an open platform for language agents in the wild with three agents, a Data Agent for analysis with Python and SQL, a Plugins Agent over 200 daily tools, and a Web Agent for autonomous browsing; dated news to 2023-10-26. publisher · read in our library knowledge/fetched/cmp_agents_openagents_readme.md · pin hfb6ef7fdb133612467e27ffa642ccbe73cfb4bd8fb0c63c1bc629ad686694e1d · accessed 2026-09-16 · source-readGrounds: Organ connector catalog
[webarena23] WebArena README (arXiv 2307.13854), fetched and read 16 Sep 2026: self-hosted realistic web environments for autonomous agents with a canonical implementation and a public leaderboard sheet; updates dated 2024-12-05 and 2024-12-20 point to TheAgentCompany for terminal and coding tasks. publisher · read in our library knowledge/fetched/cmp_automation_webarena_readme.md · pin h081c826497f9b7da4952771ee128c2215ec3532379db0fe9f7fcbd899d9c9287 · accessed 2026-09-16 · source-readGrounds: RPA UI step
[gaia23] Mialon et al. GAIA: a benchmark for General AI Assistants (arXiv 2311.12983, submitted 2023-11-21), abstract page fetched and read 16 Sep 2026: real-world questions requiring reasoning, multi-modality, web browsing and tool use, conceptually simple for humans yet hard for AI. publisher · read in our library knowledge/fetched/cmp_automation_gaia_abs.html · pin hfcebc315ea72ccad1c615e354c5218f6d34ef6ad1d5bfa0fafc15b04fa435e9e · accessed 2026-09-16 · published-paperGrounds: Organ connector catalog
[tau2bench26] tau2-bench README (Sierra Research, arXiv 2506.07982), fetched and read 16 Sep 2026: the dual-control tool-agent benchmark with a live leaderboard at taubench.com, the tau-Voice and tau-Knowledge extensions (arXiv 2603.13686 and 2603.04370, March 2026), and a v1.0.1 grading update dated July 2026. publisher · read in our library knowledge/fetched/cmp_automation_tau2bench_readme.md · pin hed39688aa5d40bd73007355df65f4e41a6c250960642e56764c3ad44da1df8d2 · accessed 2026-09-16 · source-readGrounds: Organ connector catalog
[theagentcompany24] TheAgentCompany README (arXiv 2412.14161), fetched and read 16 Sep 2026: benchmarking LLM agents on consequential real-world professional tasks in a self-hosted software company (setup scripts dated 2024-12-08) with a leaderboard on the project site; the project site itself answered a 903-byte shell over the sovereign fetcher and is not mirrored as evidence. publisher · read in our library knowledge/fetched/cmp_automation_theagentcompany_readme.md · pin hd1646f67d49d38ef1e81bf7b5057dfc3d6f172b4c8a7413a2cc78741ccb08723 · accessed 2026-09-16 · source-readGrounds: RPA UI step

generated by nx_swcompare_matrix (sovereign NishiLang organ) from knowledge/compare/automation.matrix · source checks show implementation presence; runtime and user-outcome evidence are reported separately · JavaScript supports page controls