nishi code wiki / compare / auto repair
Auto Repair Diagnostics & Repair Planning — competitive matrix v1
Published · evidence base: the car-repair SOTA brief (12 sourced claims) + the live vehreg- plane · test platforms: 2012 Sonata 2.4 (Theta II), 2013 Sienna 3.5 (2GR-FE)
Cell provenance, stated against ourselves: every competitor cell below is LITERATURE-SOURCED (July-2026 buyer's guides, vendor pages — cited in the brief), not measured by us; OURS cells cite live estate artifacts. A radar that cannot mine is a report, not an instrument — measured cells arrive when R5
nx_carbench replays documented field cases through the planner. Until then this page is honest literature, not measurement.The matrix
| Axis | ALLDATA | ProDemand | Identifix Direct-Hit | Autel Ultra S2 | Snap-on Zeus + SureTrack | NISHI |
|---|---|---|---|---|---|---|
| OEM service data mirror | Best | Yes | Partial | Partial | Partial | No (non-goal v1) |
| Labor-time estimating | Yes | Best | No | No | Partial | No (non-goal v1) |
| Aggregated confirmed fixes | Partial | Yes (SureTrack data) | Best | No | Yes | Planned R5 case bank |
| Ordered, costed, evidence-gated test PLAN | No | No | Partial (symptom trees) | Partial (guided component tests) | Partial (guided) | Planned R2 = the differentiator |
| Campaign/warranty economics INSIDE diagnosis | Partial (TSB/recall lookup, separate) | Partial | No | No | No | Planned R2 stage-0 (Theta II proves the value) |
| Failure priors from public data per platform | No | Partial | Partial (proprietary) | No | Partial (parts-frequency) | SHIPPED v0 — vehreg- 12 rows, NHTSA-derived, 2026-08-05 |
| Parts-quality rating + upgrade paths (buy-or-make) | No | No | No | No | No | Planned R3 (nobody occupies this axis) |
| Refuses verdict without measurements (BITE) | No | No | No | No | No | Planned R2 gate |
| Vehicle coverage breadth | Best | Best | Yes | Yes | Yes | 2 platforms (seed) |
| Scan hardware / bidirectional control | No | No | No | Best-value (AI inspection, topology) | Best (US-domestic + SureTrack) | No (commodity OBD adapter assumed) |
Gap queue (the build order)
| Rung | What | Tracking |
|---|---|---|
| R2 | nx_diag_plan — evidence-gated planner; campaign sweep stage-0; refuse-without-measurement BITE | debt 1785972349 |
| R3 | nx_partsgrade — grade from supersession/campaign/complaint evidence; buy-or-make upgrade map | debt 1785972349 |
| R5 | nx_carbench ruler — top-2 ≥80%, tests-to-isolation ≤3; wired before first planner promotion | debt 1785972349 |
| R4 | Automotive corpus coverage — gate ≥20/30 on the ruler query | debt 1785972413 |
| TLS | Citation re-ingest after AES-256-GCM record half ships (18 self-hosted mirrors bridge until then) | debts 1785966831 / 1785971025 |