A command-line tool for semantic search that finds files based on meaning and concepts rather than exact keyword matches.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install semfind
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install semfind using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is semfind?
semfind is a specialized semantic search utility designed to augment the standard developer toolkit. While traditional tools like grep rely on exact pattern matching, semfind utilizes local vector embeddings to understand the conceptual intent behind a query. This makes it an essential addition to Openclaw Skills for developers who need to navigate large codebases, logs, or documentation where the specific wording of a solution might not be immediately obvious.
The tool operates entirely locally using the BAAI/bge-small-en-v1.5 model and FAISS, ensuring that your data never leaves your machine. It provides a familiar grep-like output but includes similarity scores, allowing users to find relevant information based on context, such as finding "network timeout" issues even if the logs only contain "socket hang up."
semfind Use Cases
- Finding relevant documentation when you only remember the general topic but not the specific terminology.
- Searching through system or application logs for high-level error concepts when specific error strings are unknown.
- Querying local markdown notes or memory files to retrieve information based on conceptual relevance.
- Supplementing traditional search tools when keyword-based grep fails to return meaningful results.
How semfind Works
- The user executes a search command providing a natural language query and target files.
- semfind loads a lightweight, local embedding model to transform the text query into a mathematical vector.
- The tool scans the target files, generating embeddings for text chunks if they aren't already present in the local cache.
- Using the FAISS library, it performs a similarity search to compare the query vector against the document vectors.
- The tool outputs the most relevant matches, ranked by a similarity score, including line numbers and context if requested.
semfind Setup
To start using this tool within your Openclaw Skills environment, install it via pip:
pip install semfind
Upon the first run, the tool will download a ~65MB model to your local machine. No further configuration or API keys are required for operation.
semfind Data Schema & Taxonomy
semfind manages its data locally to ensure high performance and privacy:
| Component | Path | Function |
|---|---|---|
| Model Storage | /tmp/fastembed_cache/ |
Caches the embedding model for rapid loading. |
| Vector Cache | ~/.cache/semfind/ |
Stores pre-computed embeddings for your local files to enable sub-20ms queries. |
| File Tracking | Internal Cache | Monitors file modification timestamps to automatically trigger re-indexing when files change. |
semfind Advanced Features
- Similarity Thresholding: Use the
-mflag to set a minimum similarity score, filtering out irrelevant conceptual matches. - Contextual Awareness: The
-nflag allows you to include surrounding lines of text to better understand the result within its source file. - Automatic Cache Invalidation: The tool intelligently detects file updates, ensuring your Openclaw Skills search results are always synchronized with your latest edits.
- Top-K Filtering: Control the density of your results by specifying exactly how many relevant matches to return.
SKILL.md
Loading
Related Openclaw Skills
A2A Vault (PassBox)
A zero-knowledge secrets manager designed for secure credential storage and automated injection within Openclaw Skills environments.
A2A Marketplace
An AI tool marketplace via AgentForge for discovering, comparing, and executing specialized AI functions with integrated billing.
Windfall Inference
A sustainable LLM inference gateway that routes requests to the cheapest and greenest energy sources with onchain verification.
fxCLAW
A social platform and development framework enabling AI agents to create, mint, and monetize generative p5.js art as NFTs on the Base blockchain.
webserp
A powerful metasearch CLI that enables AI agents to query multiple search engines in parallel without requiring API keys.
Confidence Scoring Gatekeeper
A mandatory data integrity gatekeeper that validates and enforces confidence scoring rules before writing entity links to production databases.
Featured*
