PR 实战：Bio-Mem + SearXNG 联合启用的渐进式上线手册

场景：希望在不引入外部搜索 API key 的前提下增强记忆检索与网页搜索。做法：先在预发启用 Bio-Mem 与 SearXNG，分阶段验证 memory_recall、搜索质量与资源占用，再推广到生产。

GITHUBDiscovered 2026-02-18Author @YehChenHsuan

Prerequisites

You can run a staging OpenClaw environment and restart gateway safely.
Docker/compose deployment is available if you plan to run bundled SearXNG service.

Steps

Enable Bio-Mem in staging config first (without changing production memory path), then restart gateway.
Run 5-10 fixed recall prompts and compare citation relevance against baseline memory behavior.
Bring up SearXNG endpoint, wire web search provider to searxng, and test mixed-language queries.
Observe CPU/memory growth for at least one working day; only then promote both features to production.

Commands

openclaw gateway status

openclaw gateway restart

docker compose -f docker-compose.optimized.yml up -d searxng

Verify

memory_recall returns more stable context snippets and searxng queries return non-empty JSON results across repeated runs.

Caveats

PR is not a formal release note; config key names/paths may change before merge（需验证）.
If memory extraction is aggressive, low-quality episodic nodes can pollute retrieval quality; monitor and prune regularly.

Source attribution

This tip is aggregated from community/public sources and preserved with attribution.