Files
openclaw/docs/concepts/memory-search.md
Peter Steinberger 6fc564278d feat(agents): sessions_search full-text recall over past session transcripts (#105057)
Adds a core sessions_search agent tool backed by a SQLite FTS5 index that
lives next to the transcript rows in the per-agent database. Indexing is
transactional: user/assistant text is indexed inside the same write
transaction that persists the event, deletes drop index rows in the same
transaction, and branch rewinds mark the session dirty so the next search
rebuilds it from the same visible-path resolution sessions_history uses
(which also lazily backfills doctor-migrated databases). Search executes
gateway-side behind an additive sessions.search method with a bounded
session-key allowlist; visibility, sandbox restrictions, and snippet
redaction mirror sessions_history.

Closes #100978
2026-07-12 09:15:41 +01:00

201 lines
7.6 KiB
Markdown

---
summary: "How memory search finds relevant notes using embeddings and hybrid retrieval"
title: "Memory search"
read_when:
- You want to understand how memory_search works
- You want to choose an embedding provider
- You want to tune search quality
---
`memory_search` finds relevant notes from your memory files, even when the
wording differs from the original text. It chunks memory into small pieces and
searches them with embeddings, keywords, or both.
## Quick start
OpenClaw uses OpenAI embeddings by default. To use another provider, set it
explicitly:
```json5
{
agents: {
defaults: {
memorySearch: {
provider: "openai", // or "gemini", "voyage", "mistral", "bedrock", "local", "ollama", "lmstudio", "github-copilot", "openai-compatible"
},
},
},
}
```
`provider` can also reference a custom `models.providers.<id>` entry (for
example `ollama-5080`), as long as that entry sets `api` to `"ollama"` or
another provider id with a memory embedding adapter.
For local embeddings with no API key, install the official llama.cpp provider
plugin and set `provider: "local"`:
```bash
openclaw plugins install @openclaw/llama-cpp-provider
```
Source checkouts still need native build approval: `pnpm approve-builds`, then
`pnpm rebuild node-llama-cpp`.
Some OpenAI-compatible embedding endpoints require asymmetric `input_type`
labels, such as `"query"` for searches and `"document"`/`"passage"` for indexed
chunks. Set these with `queryInputType` and `documentInputType`; see
[Memory configuration reference](/reference/memory-config#provider-specific-config).
## Supported providers
| Provider | ID | Needs API key | Notes |
| ----------------- | ------------------- | ------------- | --------------------------------- |
| Bedrock | `bedrock` | No | Uses the AWS credential chain |
| DeepInfra | `deepinfra` | Yes | Default model `BAAI/bge-m3` |
| Gemini | `gemini` | Yes | Supports image/audio indexing |
| GitHub Copilot | `github-copilot` | No | Uses your Copilot subscription |
| Local | `local` | No | GGUF model, ~0.6 GB auto-download |
| LM Studio | `lmstudio` | No | Local/self-hosted server |
| Mistral | `mistral` | Yes | |
| Ollama | `ollama` | No | Local/self-hosted server |
| OpenAI | `openai` | Yes | Default |
| OpenAI-compatible | `openai-compatible` | Usually | Generic `/v1/embeddings` endpoint |
| Voyage | `voyage` | Yes | |
## How search works
OpenClaw runs two retrieval paths in parallel and merges the results:
```mermaid
flowchart LR
Q["Query"] --> E["Embedding"]
Q --> T["Tokenize"]
E --> VS["Vector search"]
T --> BM["BM25 search"]
VS --> M["Weighted merge"]
BM --> M
M --> R["Top results"]
```
- **Vector search** matches similar meaning ("gateway host" matches "the
machine running OpenClaw").
- **BM25 keyword search** matches exact terms (IDs, error strings, config
keys).
- **Filename search** indexes paths separately from note bodies. Exact full
paths, basenames, and filename stems rank ahead of partial path matches,
while snippets and body keyword scores still come from note content.
If only one path is available, the other runs alone.
**FTS-only mode.** Set `provider: "none"` to intentionally disable embeddings
and search with keywords only. Leaving `provider` unset or set to `"auto"`
also falls back to keyword-only ranking if no embedding auth is configured,
without erroring, and so does `provider: "local"` (the GGUF/llama.cpp
provider) when it fails.
**Explicit provider unavailable.** If you name any other provider explicitly
(for example `openai`, `ollama`, `gemini`) and it becomes unavailable at
request time (bad auth, network failure), `memory_search` reports memory as
unavailable instead of silently degrading to FTS-only results. This keeps a
broken configured provider visible. Set `provider: "none"` for deliberate
FTS-only recall, or fix the provider/auth configuration to restore semantic
ranking.
## Improving search quality
Two optional features help with a large note history.
### Temporal decay
Old notes gradually lose ranking weight so recent information surfaces first.
With the default 30-day half-life, a note from last month scores at 50% of its
original weight. `MEMORY.md` and other non-dated files under `memory/` are
evergreen and never decayed; only dated `memory/YYYY-MM-DD.md` files decay.
<Tip>
Enable this if your agent has months of daily notes and stale information
keeps outranking recent context.
</Tip>
### MMR (diversity)
Reduces redundant results. If five notes all mention the same router config,
MMR ensures the top results cover different topics instead of repeating.
<Tip>
Enable this if `memory_search` keeps returning near-duplicate snippets from
different daily notes.
</Tip>
### Enable both
```json5
{
agents: {
defaults: {
memorySearch: {
query: {
hybrid: {
mmr: { enabled: true },
temporalDecay: { enabled: true },
},
},
},
},
},
}
```
## Multimodal memory
With `gemini-embedding-2-preview`, you can index images and audio alongside
Markdown. This only applies to files under `memorySearch.extraPaths`; default
memory roots (`MEMORY.md`, `memory/*.md`) stay Markdown-only. Search queries
remain text, but they match against visual and audio content. See
[Memory configuration reference](/reference/memory-config#multimodal-memory-gemini)
for setup.
## Session memory search
For exact full-text recall from session transcripts, use [`sessions_search`](/concepts/session-search)
and then open a result with `sessions_history`. Session-memory search remains the semantic,
experimental complement.
Optionally index session transcripts so `memory_search` can recall earlier
conversations. This is opt-in: set `experimental.sessionMemory: true` and add
`"sessions"` to `sources` (default `sources` is `["memory"]`).
Session hits obey `tools.sessions.visibility`: the default `"tree"` only
exposes the current session and sessions it spawned. To recall an unrelated
same-agent session from a different session (for example a gateway-dispatched
session from a DM), widen visibility to `"agent"`.
When using the QMD backend, also set `memory.qmd.sessions.enabled: true` so
transcripts get exported into the QMD collection; `experimental.sessionMemory`
and `sources` alone do not export transcripts into QMD. See
[configuration reference](/reference/memory-config#session-memory-search-experimental).
## Troubleshooting
**No results?** Run `openclaw memory status` to check the index. If empty, run
`openclaw memory index --force`.
**Only keyword matches?** Your embedding provider may not be configured. Check
`openclaw memory status --deep`.
**Local embeddings time out?** `ollama`, `lmstudio`, and `local` use a longer
inline batch timeout by default. If the host is just slow, set
`agents.defaults.memorySearch.sync.embeddingBatchTimeoutSeconds` and rerun
`openclaw memory index --force`.
**CJK text not found?** Rebuild the FTS index with
`openclaw memory index --force`.
## Related
- [Memory overview](/concepts/memory)
- [Active memory](/concepts/active-memory)
- [Builtin memory engine](/concepts/memory-builtin)
- [Memory configuration reference](/reference/memory-config)