RAVENResearch is not verification
RAG retrieves; deep research loops; agents summarize. None of that verifies a token. An agent may retrieve, compare, and explain Raven receipts — it must not replace exact receipt verification with retrieved text, memory, summaries, screenshots, citations, or generated prose. Machine-readable: /agentic-research-boundary-policy.json.
Clarify before researching
Resolve chain, mint, requested action, evidence scope, and acceptable receipt age before expensive work. Never infer a missing mint or an ambiguous chain — ask or fail closed. If a user asks "is it safe", answer in Raven's bounded language; the word does not exist here. Prefer deterministic fields, enums, and schema inputs over model inference.
Bounded loop, state over context
The loop is: parse structured request → retrieve exact receipt → verify payloadHash, receiptId, signature, signer pin, subject, rules and freshness outside the LLM → evaluate completeness against the requested scope → retry within limits or stop with explicit gaps. Missing evidence becomes a coverage gap, never a model guess. Store large tool outputs in state with stable references. Context compression may shrink notes; it may never drop current receipt-v1 invariants (subject, rulesVersion, findings, gaps, payloadHash, receiptId, signer trust, signature status, slot, freshness). Verdict/keyId fields are legacy v2 compatibility only. Completion requires the completeness eval — not model confidence or context pressure.
Writer is not verifier
The writer gets structured verified fields plus coverage gaps plus allowed language. It may not invent evidence, omit blocking gaps, strengthen "not verified", or invent a receipt-v1 verdict. Draft reflection checks: no safety-affirming wording, no price/trading advice, findings/gaps unchanged, signer trust and freshness visible. A synthesis gap means rewrite; a research gap means retrieve within limits or state it. Citations never replace Ed25519 verification. Every material action still requires the exact receipt, verified, fresh, with gaps and policy applied. Drills 31–34 rehearse this: failure drills.