semfind
by paperboardofficial
Coding Agents & IDEsBrief
Semantic search over local text files using embeddings.
Description
What This Skill Does
semfind searches text files by meaning rather than exact pattern, using local embeddings (BAAI/bge-small-en-v1.5 + FAISS) via the terminal. It works offline with no API keys. Install with `pip install semfind`.
Covers the gap where grep returns nothing because the search term and the stored text use different words for the same concept.
When to Use It
- Finding a past error in logs when you don't remember the exact wording
- Searching project notes for a concept discussed weeks ago
- Locating relevant memory files by topic rather than keyword
- Recovering a decision or rationale buried in documentation
- Scanning multiple markdown files for semantically related content
Example Workflow
User asks: find any notes about database connection problems in memory files
- 1
Run grep first: grep "database connection" memory/*.md
- 2
Grep returns no results
- 3
Run semfind: semfind "database connection problems" memory/*.md -k 5
- 4
Review results with similarity scores to identify relevant files
- 5
Return matched lines and their similarity scores to the user
memory/2026-01.md:42: Postgres refused connections after restart (0.831)
Installation
clawhub install paperboardofficial/semfindRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.