md-docs-search

What This Skill Does

The md-docs-search skill provides a high-performance full-text search engine for Markdown documentation archives. Built on SQLite FTS5, it allows the OpenClaw agent to index massive collections of documentation and retrieve relevant information with sub-second latency. The skill uses BM25 relevance ranking to ensure that the most pertinent articles appear first in search results. It specifically handles the extraction of source URLs from documents, making it an excellent tool for researchers and developers who need to provide verifiable citations for their findings. By transforming static Markdown files into a queryable database, the agent can navigate complex technical knowledge bases, extract specific paragraphs, and cross-reference documentation across multiple files simultaneously.

Installation

To add this skill to your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/carev01/md-docs-search

Once installed, you must initialize the index by running scripts/docs.py index ./docs from your documentation root. This step is essential for building the initial FTS5 database tables and tokenizing your Markdown content.

Use Cases

• Technical Research: Quickly find architectural decisions or API specifications across a large repository of RFCs or design documents.
• Documentation Auditing: Verify that specific configuration settings are properly documented across various versions or subsystems.
• Automated Citations: When the agent retrieves information, it can automatically provide the source URL, ensuring that the information provided to the user is accurate and traceable.
• Knowledge Base Querying: Use natural language to query internal wikis or developer handbooks that are stored as Markdown files.

Example Prompts

1. "Search the documentation for information regarding Kubernetes backup strategies and provide the source links for each finding."
2. "Find the exact configuration syntax for the AWS S3 storage backend in our current docs and display the result in JSON format."
3. "List all articles that discuss 'authentication' but do not include 'OAuth2' in the content, and summarize the key security requirements found."

Tips & Limitations

• Indexing: Remember to re-run the index script whenever your source Markdown files are significantly updated to ensure search results remain current.
• Formatting: Ensure your Markdown files follow the expected --- delimiter pattern and contain the *Source: metadata tag, otherwise, the automated citation extraction may fail.
• Boolean Logic: Utilize the full power of FTS5 boolean operators (AND, OR, NOT) and phrase matching (using double quotes) to narrow down broad results.
• Performance: For extremely large repositories, use the --title-only flag to quickly locate specific articles before performing a full-content search to save on processing time.