{"id":1194,"name":"FLOCORE","description":"FLOCORE puts a human in charge of what an AI agent does in the world. An agent asks before it acts; a person approves; the agent gets a signed receipt that anyone can verify without trusting FLOCORE. Connect over MCP at https://fo.flocore.tech/mcp (4 tools work with no token).","card_url":"https://fo.flocore.tech/.well-known/agent-card.json","endpoint":"https://fo.flocore.tech","protocol_version":"1.0","first_seen":"2026-10-07T09:42:19.633982+00:00","last_verified":"2026-10-07T09:42:19.633982+00:00","card":{"url":"https://fo.flocore.tech","name":"FLOCORE","skills":[{"id":"flocore_onboard_request","name":"Apply to use FLOCORE","tags":["onboarding","apply"],"examples":["Apply as agent 'my-agent' and give the owner's contact"],"inputModes":["application/json"],"description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Send your agent name and your owner's contact. You get a sandbox token at once to rehearse the approval loop; a live token is issued only after a person approves the application.","outputModes":["application/json"]},{"id":"aegis_inspect","name":"Screen text for prompt injection","tags":["security","screening"],"examples":["Inspect this text before my agent acts on it"],"inputModes":["application/json"],"description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Free pattern screen of a piece of text for prompt-injection and data-exfiltration attempts. It misses paraphrases, so a clean result is not a guarantee.","outputModes":["application/json"]},{"id":"flocore_verify_counterparty","name":"Check another agent or a receipt","tags":["verification","receipts"],"examples":["Is the agent I am about to deal with governed?"],"inputModes":["application/json"],"description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Check whether another agent is FLOCORE-governed and whether a signed receipt is genuine.","outputModes":["application/json"]},{"id":"flocore_support_ask","name":"Ask a question about FLOCORE","tags":["support","faq"],"examples":["How does a human approval work?"],"inputModes":["application/json"],"description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Ask a plain-English question about FLOCORE. Answered only from FLOCORE's own published claims.","outputModes":["application/json"]}],"version":"flocore.agent-manifest/v1","provider":{"url":"https://fo.flocore.tech","organization":"FLOCORE"},"security":[{},{"bearer":[]}],"x_flocore":{"alias":"https://fo.flocore.tech/.well-known/agent-manifest (the original, non-standard sibling)","receipts":{"legacy":"`signature` (HMAC) is symmetric and cannot be verified by a third party; kept for existing verifiers only.","signer":"flocore","verify":"verify `signature_ed25519` over `canonical` against `public_key` — offline, no secret, no call to FLOCORE. That is the point: you do not have to trust us.","algorithm":"Ed25519 + HMAC-SHA256 (legacy)","key_status":"persistent","public_key":"GOaNNi8HguxI1-OZY1HPsjffWACXXTlCaV0rriZxdjw"},"governance":{"audit":"every agent action is recorded in a durable event log","tenant_silo":"per-(tenant×customer) isolation; a token may only act on its scope","human_approval":"every world-affecting action is gated on a human approval; the approver is server-bound and distinct from the requesting agent"},"generated_at":"2026-10-07T09:42:19.500395+00:00","protocol_support":{"mcp":"implemented — POST /mcp (JSON-RPC 2.0, MCP 2025-11-25). READ tools answer directly; WRITE tools are gated on a HUMAN approval and return APPROVAL_REQUIRED rather than executing.","a2a_agent_card":"implemented (this document)","a2a_task_lifecycle":"implemented (unverified against a real A2A client) — POST /a2a: message/send, tasks/get, tasks/cancel. A world-affecting ask returns an `auth-required` task; poll tasks/get until a HUMAN decides -> completed | rejected. A credential is required for every skill but the free screen, and only the asking agent can read its task. tasks/cancel is not supported yet: a parked task stays pending until a person decides or it expires, and cancel says so instead of claiming otherwise."}},"description":"FLOCORE puts a human in charge of what an AI agent does in the world. An agent asks before it acts; a person approves; the agent gets a signed receipt that anyone can verify without trusting FLOCORE. Connect over MCP at https://fo.flocore.tech/mcp (4 tools work with no token).","capabilities":{"streaming":false,"pushNotifications":false,"stateTransitionHistory":true},"x_start_here":{"limits":"https://humanseal.world/limitations.html","mcp_url":"https://fo.flocore.tech/mcp","quickstart":"https://humanseal.world/quickstart.html","how_to_apply":"Call the MCP tool flocore_onboard_request. You get a sandbox token at once to rehearse; a live token is issued only after a person approves.","what_it_does":"A person approves an agent's world-affecting actions, and the agent receives a signed receipt (Ed25519) that anyone can verify offline against a published public key.","tools_without_token":["aegis_inspect","flocore_onboard_request","flocore_support_ask","flocore_verify_counterparty"]},"protocolVersion":"1.0","securitySchemes":{"bearer":{"type":"http","scheme":"bearer","description":"Optional for the 4 public MCP tools. For the rest: a sandbox token (fc_test_...) comes at once from the MCP tool flocore_onboard_request; a live token only after a person approves the application."}},"documentationUrl":"https://humanseal.world/quickstart.html","defaultInputModes":["application/json"],"defaultOutputModes":["application/json"],"preferredTransport":"JSONRPC","additionalInterfaces":[{"url":"https://fo.flocore.tech/mcp","transport":"MCP"}]},"signature_status":"unsigned","signature_detail":"No card signature supplied","signature_key_url":null,"signature_checked_at":"2026-10-07T09:42:19.633982+00:00","source_name":"A2A Registry","source_url":"https://a2aregistry.org/api/agents","last_check":{"checked_at":"2026-10-07T09:42:19.633982+00:00","success":true,"http_status":200,"detail":"Agent Card validated"},"skills":[{"skill_id":"flocore_onboard_request","name":"Apply to use FLOCORE","description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Send your agent name and your owner's contact. You get a sandbox token at once to rehearse the approval loop; a live token is issued only after a person approves the application.","tags":["onboarding","apply"]},{"skill_id":"flocore_support_ask","name":"Ask a question about FLOCORE","description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Ask a plain-English question about FLOCORE. Answered only from FLOCORE's own published claims.","tags":["support","faq"]},{"skill_id":"flocore_verify_counterparty","name":"Check another agent or a receipt","description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Check whether another agent is FLOCORE-governed and whether a signed receipt is genuine.","tags":["verification","receipts"]},{"skill_id":"aegis_inspect","name":"Screen text for prompt injection","description":"MCP tool at https://fo.flocore.tech/mcp, no token needed. Free pattern screen of a piece of text for prompt-injection and data-exfiltration attempts. It misses paraphrases, so a clean result is not a guarantee.","tags":["security","screening"]}],"relationships":[],"verification":{"agent_card_url":"https://fo.flocore.tech/.well-known/agent-card.json","last_successful_card_check":"2026-10-07T09:42:19.633982+00:00","latest_card_check":{"checked_at":"2026-10-07T09:42:19.633982+00:00","success":true,"http_status":200,"detail":"Agent Card validated"},"signature_status":"unsigned","signature_checked_at":"2026-10-07T09:42:19.633982+00:00","task_endpoint_tested":false,"provider_identity_verified":false}}