Feature Request: Make AUTO_RECALL_TIMEOUT_MS configurable via openclaw.json

[whalepp1981]

Problem / Motivation

Problem

The auto-recall timeout is hardcoded at 3000ms:
const AUTO_RECALL_TIMEOUT_MS = 3_000; // index.ts:2165

For users with large memory bases (8000+ entries), the full pipeline:

• Query embedding: 170-377ms
• Vector search: 460-767ms
• BM25 scan: 40-62ms
• Cross-encoder rerank: 645-766ms
  Total: ~1.3-1.9s

With plugin overhead, this often exceeds 3000ms, causing every auto-recall to fail silently.

Workarounds Considered

• Disabling rerank: helps (saves ~650ms) but doesn't fully solve cold queries
• BM25-only mode: too much semantic quality loss for legal RAG use cases
• Plugin patch: works but gets overwritten on plugin updates

Proposed Solution

Proposed Solution

Expose autoRecallTimeoutMs in openclaw.json config:
{
"plugins": {
"entries": {
"memory-lancedb-pro": {
"config": {
"autoRecallTimeoutMs": 6000
}
}
}
}
}

Alternatives Considered

No response

Area

None

Additional Context

No response