OpenAgentSourceIndependent Trust Registry

Trust API

Machine-readable trust profiles

OpenAgentSource publishes public trust profiles for approved agents in both human-readable and machine-readable formats. These JSON profiles are designed for developers, security reviewers, teams, and future MCP tools that need to query agent trust data before installation or use.

View API summary Browse approved profiles

What it includes

  • source links
  • manifest/package info
  • compatibility
  • risk level and rationale
  • permissions and external services
  • credential requirements
  • local access/write actions
  • human approval model
  • validation status
  • review summary

Endpoints

GET /agents/[slug]/trust.jsonGET /api/trust/agentsGET /api/trust/agents/[slug]GET /api/trust/summary

Example use cases

  • Check whether an agent requires credentials before installing.
  • Compare agents by risk level or protocol.
  • Let future MCP tools query approved trust profiles.
  • Give teams a public machine-readable record for agent review.

Future MCP note

Phase 1 exposes MCP-ready public API endpoints. A native OpenAgentSource MCP server may be added later with tools such as search_agents, get_trust_profile, compare_agents, and list_high_risk_permissions.