Agents

Endpoints for managing agent trust levels and resolving cross-agent memory conflicts.

Base URL: http://localhost:3050

All request bodies are JSON (Content-Type: application/json). Field names on the raw HTTP prototype surface use snake_case.

PUT /v1/agents/trust

Set the trust level for an agent.

Request:

{
  "agent_id": "agent-gpt4",
  "user_id": "ethan",
  "trust_level": 0.85,
  "display_name": "GPT-4 Assistant"
}

Field	Type	Required	Notes
agent_id	string	yes
user_id	string	yes
trust_level	number	yes	0.0–1.0
display_name	string	no	Human-readable agent name

Response:

{ "agent_id": "agent-gpt4", "trust_level": 0.85 }

GET /v1/agents/trust

Get trust level for an agent.

Query params: agent_id (required), user_id (required)

Response:

{ "agent_id": "agent-gpt4", "trust_level": 0.85 }

Example:

curl 'http://localhost:3050/v1/agents/trust?agent_id=agent-gpt4&user_id=ethan'

GET /v1/agents/conflicts

List open cross-agent memory conflicts.

Query params: user_id (required)

Response:

{
  "conflicts": [
    { "id": "conflict-1", "agent_a": "agent-gpt4", "agent_b": "agent-claude", "status": "open" }
  ],
  "count": 1
}

PUT /v1/agents/conflicts/:id/resolve

Manually resolve a conflict.

Request:

{ "resolution": "resolved_new" }

resolution must be one of: resolved_new, resolved_existing, resolved_both.

Response:

{ "id": "conflict-1", "status": "resolved_new" }

POST /v1/agents/conflicts/auto-resolve

Auto-resolve expired conflicts for a user.

Request:

{ "user_id": "ethan" }

Response:

{ "resolved": 3 }