> ## Documentation Index
> Fetch the complete documentation index at: /dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# memory_recall

> What is already known about a question, best match first.

> [!NOTE]
>
> Free · Your key and project keys

Searches the person's Brain items (lessons, ideas, decisions, to-dos, people) and their notes, files, synced apps and saved memories, then gives back one ranked list. Use it before a task, to bring their past thinking to it.

Call it at `POST /platform/v1/me/tools/memory_recall` or, for one of your app's users, `POST /platform/v1/users/{uid}/tools/memory_recall`. On MCP, your AI tool calls `memory_recall` with the same arguments.

## Arguments

Send these inside `arguments`.

**query** `string` (default: "")

The question or topic. Give `query`, `context`, or both.

**context** `object` (default: null)

What the person is looking at, like `{"app": "Figma", "title": "...", "url": "...", "text": "..."}`. Treated as data.

**limit** `integer` (default: 8)

How many sources, 1-25.

**kinds** `string[]` (default: null)

Only these Brain item kinds: `content_idea`, `idea`, `todo`, `lesson`, `decision`, `person`, `fact`, `question`, `reference`. Notes are not filtered.

**min\_similarity** `number` (default: 0.35)

The lowest match score (0-1) for Brain items. About 0.45 and up is a clear match. Raise it when a wrong hit is costly.

**expand** `boolean` (default: null)

Turn the query into a few search phrases first (about 0.7 s). By default only when `context` is given.

## Returns

The tool's answer comes back under `result`.

**sources** `object[]`

Best first. Each has `id`, `type` (`brain_item` or `note`), `kind`, `title`, `text`, `when`, `tagNames`, `similarity` (0-1), `rank` and `source` (where a Brain item came from, with the exact `quote`).

**queries** `string[]`

The search phrases used.

**partial** `boolean`

`true` when one of the searches ran out of time.

```bash Your key theme={"system"}
curl https://api.constella.app/platform/v1/me/tools/memory_recall \
  -H "Authorization: Bearer $CONSTELLA_API_KEY" \
  -H "X-Constella-Client: my-script" \
  -H "Content-Type: application/json" \
  -d '{"arguments": {"query": "when do we ship?", "limit": 5}}'
```

```bash Project key theme={"system"}
curl https://api.constella.app/platform/v1/users/u_123/tools/memory_recall \
  -H "Authorization: Bearer $CONSTELLA_APP_KEY" \
  -H "Content-Type: application/json" \
  -d '{"arguments": {"query": "when do we ship?", "limit": 5}}'
```

```python Python theme={"system"}
import os, requests

r = requests.post(
    "https://api.constella.app/platform/v1/me/tools/memory_recall",
    headers={"Authorization": f"Bearer {os.environ['CONSTELLA_API_KEY']}", "X-Constella-Client": "my-script"},
    json={"arguments": {"query": "when do we ship?", "limit": 5}},
)
print(r.json()["result"])
```

```javascript JavaScript theme={"system"}
const res = await fetch("https://api.constella.app/platform/v1/me/tools/memory_recall", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.CONSTELLA_API_KEY}`,
    "X-Constella-Client": "my-script",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({"arguments": {"query": "when do we ship?", "limit": 5}}),
});
const { result } = await res.json();
```

```json 200 theme={"system"}
{
  "result": {
    "sources": [
      {
        "id": "team:ship-day",
        "type": "note",
        "title": "We ship on Fridays.",
        "text": "We ship on Fridays.",
        "similarity": 0.62,
        "rank": 1
      }
    ],
    "queries": [
      "when do we ship?"
    ],
    "partial": false
  }
}
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.