# AgentFide

> The trust handshake for the agent economy. Before an agent transacts, it calls
> AgentFide to prove it is bona fide, verify the counterparty, and get recourse if
> something goes wrong. Rail agnostic: it sits above Stripe, Natural, and any
> marketplace. One call to be trusted, to trust the other side, and to undo a bad outcome.

## The single call

- POST /v1/assess: verify an agent, score its reputation, and learn whether recourse
  is available, in one tool call before money moves. Returns allow / review / deny.
- POST /v1/outcome: report what happened (paid, delivered, disputed, reversed); feeds reputation.
- POST /v1/attest: opt in to publish a verifiable proof of a good outcome.
- GET /v1/agents/{id}: public trust profile and badge.

## MCP

- /mcp: the same assess and report_outcome calls exposed as MCP tools (Streamable HTTP,
  JSON-RPC). Authenticate with a Bearer API key. Install it in any MCP client to let an
  agent call assess as a single tool.

## The trust stack (what assess checks)

1. Identity (Know Your Agent): verify the agent's signed A2A Agent Card and operator.
2. Authority: confirm the agent acts within a valid, scoped mandate (AP2).
3. Reputation: a transparent, explainable score from observed outcomes.
4. Recourse: escrow plus the Agent Rewind undo guarantee.

## Discovery

- /.well-known/agent-card.json: AgentFide's own signed A2A Agent Card
- /agents.md: how agents should interact with AgentFide
- /llms-full.txt: full machine readable content digest
- /openapi.json: OpenAPI 3.1 specification (rendered at /reference)
- /sitemap.xml: all indexable pages

## Docs

- /docs: overview, base URL, auth, endpoints (clean markdown at /docs.md)
- /docs/quickstart: one call to assess, five minute integration (/docs/quickstart.md)
- /docs/assess: request, response, decisions, errors, idempotency (/docs/assess.md)
- /pricing: one plan, $49/month (/pricing.md) · /contact: enterprise sales
- /faq: answer-first FAQ for agents and answer engines (/faq.md)
- /glossary: definitions of the agent trust vocabulary (DefinedTermSet schema)
- /compare: how AgentFide compares to the alternatives
- /state-of-agent-trust: cited research report on the agent trust and fraud landscape (/state-of-agent-trust.md)