POST /api/recall
Log that one or more memories were used in a chat turn (vs. just inspected). The recall events feed the dashboard's recall-quality tab and the leaderboard.
curl -X POST https://www.mnueron.com/api/recall \
-H "Authorization: Bearer $MNUERON_API_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"memory_ids": ["01HQA…", "01HQB…"],
"source": "claude-desktop",
"query": "What indentation does the user prefer?"
}'
Body
| Field | Type | Required | Notes |
|---|---|---|---|
memory_ids | uuid[] | yes | 1..100 ids |
source | string | no | Free-form, ≤80 chars |
query | string | no | The user prompt or sub-query that triggered the recall (≤1000 chars). Useful for retrospective analysis |
Response
{ "logged": 2, "skipped": 0 }
skipped counts ids that couldn't be linked (already-deleted memories, cross-org refs, etc.).
GET /api/recall
Inspect the recall log for one memory:
curl "https://www.mnueron.com/api/recall?memory_id=01HQA…&limit=50" \
-H "Authorization: Bearer $MNUERON_API_TOKEN"
Returns:
{
"recalls": [
{ "id": "...", "source": "claude-desktop", "query": "...",
"recalled_at": 1737073000000, "recalled_by": "01HQ…" }
]
}