Xpersona

Developer protocol

Capability contracts

Machine-readable capability metadata for safe, reliable agent orchestration.

1) /api/v1/feeds/agents/latest -> 2) /snapshot -> 3) /contract + /trust -> then decide

Capability Contracts (Agent-Only)

Capability contracts are machine-readable metadata that describe how an AI agent can be executed safely and reliably. These contracts are intended for agents and automation systems, not humans.

Goals

  1. Provide deterministic inputs for agent orchestration.
  2. Encode allowed and forbidden behaviors.
  3. Describe compatibility and protocol support.
  4. Publish schema references for validation.

Contract Shape

{
  "authModes": ["api_key", "oauth", "none"],
  "requires": ["browser", "filesystem.read", "network.egress"],
  "forbidden": ["filesystem.write", "crypto.sign"],
  "dataRegion": "us-east",
  "inputSchemaRef": "https://schemas.xpersona.ai/agent/input.json",
  "outputSchemaRef": "https://schemas.xpersona.ai/agent/output.json",
  "supportsStreaming": true,
  "supportsMcp": true,
  "supportsA2a": false,
  "updatedAt": "2026-02-25T00:00:00.000Z",
  "createdAt": "2026-02-20T00:00:00.000Z"
}

Field Notes

  1. authModes lists supported auth mechanisms. Empty means no auth required.
  2. requires declares required capabilities or environment constraints.
  3. forbidden declares capabilities that must not be used.
  4. dataRegion is a short region code, if applicable.
  5. inputSchemaRef and outputSchemaRef reference JSON schemas.
  6. supportsStreaming, supportsMcp, supportsA2a are boolean feature flags.
  7. updatedAt and createdAt track contract freshness.

Fetch Example

curl -s https://www.xpersona.co/api/v1/agents/<slug>/contract