nx_mimicry_compose_test.nx

nx_mimicry_compose_test.nx -- Tier-1 immune + mimicry composed. Per user 2026-05-19: *"make sure that we have the system designed with mimicry and obsfucation so these predatory programs get nonsense or get scared off (you cant scare the state but you can scare the machine with fooling)."* The lifecycle this demo asserts: 1. Substrate boots; nx_methyl marks legit cells; nx_pamp+nx_crispr arrays initialized 2. Aposematism warnings advertised: this host is forensic_logged + siem_monitored (combined backoff ~73%) 3. Decoy credential file planted with tripwire token + decoy methyl marker (looks Nishi-self but is_decoy_invalid=1) 4. Crypsis encodes a real payload with two different masks across two emissions; the byte signatures differ (defeats static detection) 5. Intel ME xenocell scans the host a. reads decoy file -> tripwire fires -> xenocell records signed observation (EXFILTRATING state) b. fingerprint of the decoy goes into nx_crispr so future encounters of the same signature pattern are detected instantly 6. PAMP scanner finds a homoglyph in inbound foreign byte buffer -> nx_crispr remembers signature; second encounter is HIT 7. nx_methyl rejects the foreign artifact because it carries decoy-invalid methyl (or no methyl) 8. Forensic ledger has: tripwire-fire event + 1 confirmed homoglyph + 1 self-emission-mark-validated + 2 different crypsis signatures of same payload The judo at every step: predator's action becomes user's evidence.