Tools reference
The acf-mcp server exposes twelve MCP tools — seven REASON (deterministic reasoning signed against the doctrine) and five READ (corpus resource lookup). No tool calls an internal LLM; every output embeds doctrine_signature, doctrine_hash, doctrine_version and generated_at.
doctrine_version and you get exactly the same output (modulo the timestamp). That is what makes every reply verifiable offline against the ACF® Ed25519 public key.REASON tools ×7
Each REASON tool produces a structured qualification from a declared input. The output is signed and explicitly positioned as preliminary — requires_human_review: true is always set.
acf.advisorREASONGovernance assessment
Turn a free-text use case into a structured ACF® assessment: principles, autonomy, fiches, obligations, first actions.
acf.classify-agentREASONPre-go-live qualification
Ten qualified-enum fields → one structured pre-flight: autonomy level, criticality, AI Act role, controls, sign-offs.
acf.assess-autonomyREASONAutonomy level
Should the agent propose, decide, or execute? Get the N0–N3 recommendation with go/no-go, gating, and kill switch design.
acf.identify-governance-gapsREASONGap analysis
What does ACF® say you should have that you don't? Gap analysis across 6 maturity dimensions, prioritised remediations.
acf.map-ai-act-obligationsREASONAI Act mapping
Exhaustive AI Act obligations for a qualified-high-risk system, ventilated by lifecycle phase with deadlines.
acf.assign-ddao-controlsREASONDDAO controls
Concrete DDAO controls scoped to the agent’s autonomy and risk: kill switches, escalation, audit trails, ownership.
acf.evaluate-agent-mandateREASONMandate audit
Audit an existing agent mandate against the ACF® reference: strengths, gaps, required additions for sign-off.
READ tools ×5
READ tools give direct access to the ACF® corpus: methodological cards, regulator articles, glossary. Ideal for an agent to cite ACF® precisely in a user reply.
acf.searchREADCorpus search
Search the full ACF® corpus and rank resources by relevance to the question at hand.
acf.fiche.lookupREADFiche lookup
Retrieve the canonical ACF® methodological card for a specific governance question, with example and related cards.
acf.regulation.articleREADRegulator article
Verified article text (AI Act, GDPR, DORA, NIS2, ISO 42001) + its operational translation into ACF® principles and fiches.
acf.glossary.defineREADTerm definition
Canonical ACF® definitions: DDAO, Decision Sovereignty, autonomy transitions, kill switch, and the rest of the doctrine.
acf.citeREADCitation
Properly formatted citation of ACF® in APA, MLA, Chicago, ISO 690 or BibTeX style for theses, filings, board memos.
The shared signed footer
Whichever tool, the output embeds the same traceability fields — this is ACF®'s audit contract.
doctrine_version— semantic version of the doctrine at call timedoctrine_hash— sha256 of the signed contentdoctrine_signature— ed25519 over the hash, verifiable offline with the public keydoctrine_archive_url— permanent archive URL of the doctrine usedregulatory_snapshot— regulator corpus the tool reasoned againstgenerated_at— ISO 8601 timestamprequires_human_review: true— constant; no tool substitutes for a human governance decision