Multi-agent endpoints

These endpoints require the x-mneva-agent header to attribute observations. The MCP wrapper sends it automatically from your MNEVA_AGENT env var; in REST you set it yourself. See Multi-Agent.

POST /v1/note_about

Record an outside view of another agent on this brain. See note_about.

curl -X POST https://mneva.dev/v1/note_about \
  -H "x-mneva-key: $MNEVA_TEAM_KEY" \
  -H "x-mneva-agent: alice" \
  -H "content-type: application/json" \
  -d '{
    "subject_agent": "bob",
    "observation": "<what you observed>",
    "kind": "pattern"
  }'

Field	Type	Required	Notes
`subject_agent`	string	yes	The agent observed. Can be your own name (self-observation).
`observation`	string	yes	What you observed.
`kind`	`characterization` \| `feedback` \| `pattern`	no	Optional classification.

Response:

{ "id": 1, "observer_agent": "alice", "subject_agent": "bob" }

observer_agent is taken from the x-mneva-agent header. If absent, it defaults to default.

GET /v1/outside_views

Recall observations. See outside_views.

curl "https://mneva.dev/v1/outside_views?subject_agent=alice&limit=10" \
  -H "x-mneva-key: $MNEVA_TEAM_KEY"

Query param	Type	Default
`subject_agent`	string	(none — returns all subjects)
`limit`	int 1..100	20

Response:

{
  "observations": [
    {
      "id": 1,
      "observer_agent": "alice",
      "subject_agent": "bob",
      "observation": "...",
      "kind": "pattern",
      "created_at": "..."
    }
  ]
}

Newest first. Without subject_agent, returns the most recent observations across all subjects on this brain.