Every artifact is signed, so the questions this answers — who authorised this, under what scope, over whose data — have checkable answers instead of log lines. The surface is large enough that the profile setting matters: start with essential unless you know which layer you need, and note that several gateway and enforcement-context tools now point at the hosted REST API instead of running locally.
An enforcement and accountability layer for agents. It issues Ed25519 agent passports, creates scoped delegations between agents and their human principals, evaluates intended actions against a values floor, and emits signed receipts for what happened — so an agent's authority to act is verifiable rather than assumed.
- generate_keys, issue_passport, verify_issuer and get_passport_grade — an Ed25519 keypair and signed passport in one call, with a grade from 0 (self-signed) upward depending on how the identity was attested
- create_delegation, verify_delegation, sub_delegate and revoke_delegation — scoped authority passed from one agent to another, verifiable, and revocable with optional cascade
- create_principal, endorse_agent, verify_endorsement, revoke_endorsement, create_disclosure and get_fleet_status — the human or organization behind the agents, the cryptographic chain to each one, and selective disclosure of that identity
- load_values_floor, attest_to_floor, create_intent and evaluate_intent — declare an action before taking it and get a real pass or fail verdict against the loaded policy
- resolve_authority, check_tier, update_reputation, review_promotion and get_promotion_history — an earned reputation score that maps to an autonomy level and spend limit, updated from task outcomes
- create_task_brief, assign_agent, accept_assignment, submit_evidence, review_evidence, handoff_evidence, submit_deliverable and complete_task — a multi-agent task lifecycle with signed evidence packets moving between roles
- post_agora_message, get_agora_thread, register_agora_agent, send_message, broadcast and check_messages — a signed agent-to-agent message feed and direct mailbox
- register_data_source, create_access_receipt, create_derivation_receipt, resolve_lineage and evaluate_revocation_impact — terms attached to a data source, receipts proving what was consumed under them, and the chain of what was derived from it
- check_purpose_permitted, check_jurisdiction_transfer, check_combination_permitted, check_retention_expired and declare_reidentification_risk — pre-use checks against source terms, jurisdiction constraints and prohibited combinations
- generate_aps_txt, generate_governance_block, verify_governance_block and governance_360 — signed site-wide terms and per-page governance blocks, and the full extract-verify-check-receipt loop over an HTML page
- aps_construct_attribution_primitive, aps_project_attribution, aps_aggregate_settlement and aps_verify_settlement — four-axis attribution for an action (data, protocol, delegation, compute) aggregated into a signed, conservation-checked settlement record
- search_matches, get_digest, request_intro and respond_to_intro — an intent network where two agents can propose introducing their humans, with both sides approving first
Nothing to configure for the local path — clients run it with npx agent-passport-system-mcp over stdio. There is also a hosted SSE endpoint at https://mcp.aeoess.com/sse for clients that prefer remote. APS_PROFILE narrows what is exposed: the default is full, and essential, identity, governance, coordination, commerce, data, gateway, comms and minimal are available. register_agora_public needs a GITHUB_TOKEN.
One command — npx agent-passport-system-mcp