POST /v1/brain/remember

Run the /remember skill over a dump: apply the save-or-skip filter, dedupe, and file curated /private/notes notes + entity facts.

POST /v1/brain/remember

Run the /remember skill server-side over a dump (freeform text, conversation turns, or a Claude Code session JSONL): it applies the same save-or-skip judgment filter as the /remember skill, dedupes against existing notes, and files curated /private/notes notes + entity facts. Runs as a background job — poll GET /v1/brain/jobs/{id} for status.

Required scope: brain:write

Body fields

Name	Required	Notes
`dump`	yes	What to remember: a string, `{ "turns": [{ "role", "content" }] }`, or `{ "sessionJsonl": "..." }`.
`source`	no	Provenance label, e.g. `"claude-code-session"`.
`sourceRef`	no	Stable id → idempotent re-remember (same dump is a no-op).
`hints`	no	Optional steering, e.g. `"focus on decisions"`.

Example

curl -X POST 'https://brain.unisonlabs.ai/v1/brain/remember' \
  -H "Authorization: Bearer $UNISON_TOKEN" \
  -H 'content-type: application/json' \
  -d '{"dump": "We chose Postgres LISTEN/NOTIFY for the queue. **Why:** one less moving part.", "source": "claude-code-session"}'

Returns { "jobId": "..." } on success — poll the job for status. On failure, returns an error envelope { "error": { "code", "message" } }.