Semantic Memory Search with memsearch
Add vector-powered semantic search on top of OpenClaw's plain-markdown memory files with memsearch, so you can instantly find any past memory by meaning instead of grepping keywords or loading whole files into context.
Requirements
- Python 3.10+ with pip or uv.
How to use
Works with any embedding provider: OpenAI, Google, Voyage, Ollama, or fully local with no API key needed.
No OpenClaw skills are required, because memsearch is a standalone Python CLI/library.
Markdown stays the source of truth: the vector index is just a derived cache that you can rebuild anytime with memsearch index, and your memory files are never modified.
- Install memsearch:
pip install memsearch
- Run the interactive config wizard:
memsearch config init
- Index your OpenClaw memory directory:
memsearch index ~/path/to/your/memory/
- Search your memories by meaning:
memsearch search "what caching solution did we pick?"
- For live sync, start the file watcher — it auto-indexes on every file change:
memsearch watch ~/path/to/your/memory/
- For a fully local setup (no API keys), install the local embedding provider:
pip install "memsearch[local]"
memsearch config set embedding.provider local
memsearch index ~/path/to/your/memory/
Links
Source: hesamsheikh/awesome-openclaw-usecases and its contributors. License: MIT.
Copyright (c) 2026
Translation and editorial adaptation by ClawCases.
>Resources
>Related Use Cases
Connect YantrikDB Cognitive Memory to Hermes via MCP
Connect the YantrikDB cognitive memory engine to Hermes through its built-in MCP client, giving the agent tool-call operations to store, recall, relate, and consolidate facts.
Document to Action Items
Extract cited obligations, deadlines, and tasks from documents into structured facts and proposed actions, keeping low-confidence OCR and ambiguous language visible; extraction is not legal advice.
Team Telegram Assistant with Hermes
Run a shared, always-on Hermes assistant for a small team over Telegram, giving each authorized user private sessions with their own memory for coding, research, and ops.