| Role | Status | What it actually does |
|---|
MENTOR | conectado | The only role a normal zaynor chat/serve run actually invokes; explains an already-sealed result, never re-invokes the engine. |
INVESTIGATOR | implementado, sin invocación automática | Collects a bounded evidence window and re-verifies custody hashes; real and tested, no CLI/API command calls it yet. |
FLEET_COMMANDER | implementado, sin invocación automática | Writes to the investigation log; real and tested, no caller wired yet. |
DETECTION_ENGINEER | implementado, sin invocación automática | Drafts a candidate detection rule anchored to a real sealed finding; real and tested, not wired to a command yet. |
DISPATCHER | implementado, sin invocación automática | Catalog of evidence types the engine can actually analyze; real and tested, not wired to a command yet. |
ENDPOINT_HUNTER / PERSISTENCE_HUNTER | out of scope | Would need a live EDR collection backend this project does not have. |
THREAT_INTEL | out of scope, for now | A portable VirusTotal/GTI enrichment exists but is not wired in — external network dependency, pending decision. |
Findings themselves come from ZAYNOR's own deterministic Mode 1 engine, not from a named agent above — no per-finding agent attribution exists in the sealed contract (AuthoritativeFinding has no agent field). This table names what ZAYNOR's own local, Ollama-driven agent layer does around that sealed result.