vector_search
Semantic search over notes. Falls back to keyword search.
Uses Google embeddings when a Google key is set. Otherwise it searches by keyword and says so.
- Tool name
- vector_search
- Group
- Memory
- Risk
- low
- Needs approval
- never
- Approval rule
- Read only. Sends note text to Google when embeddings are on.
- Allowed in scheduled jobs
- Yes
- Default
- on
Parameters
| Name | Type | Required | Default | Meaning | Limits |
|---|---|---|---|---|---|
| query | string | Yes | Words to look for. | ||
| limit | integer | No | 5 | Maximum number of results. |
Returns
Ranked matches.
Example (made up for this page)
{
"query": "demo"
}1. global/deploy-notes.md (score 0.82)Side effects
- Sends text to the embedding service
- Writes embedding files in the background.
How it can fail
| When | What the agent sees |
|---|---|
| No Google key | A keyword result with a note that vectors are off |
See also:Memory tools
Checked against the source at commit f4fd89f on 29 Oct 2026. Everything here is merged on the main branch; no release has been tagged yet. The source is private, so these pages are the public description.