bmb-skills-hub

To generate an index.json file for a multi-skill repository, run the repository's indexing script to scan your skills directory and output a machine-readable catalog of available skills.

Semantic search for agent-ready skills works by generating an index.embeddings.json file from your skill metadata, allowing agents to retrieve relevant skills across sessions based on meaning rather than exact keyword matches.

You can create an mcp-manifest.json file by running the skill catalog generator, which exposes your skills so they can be discovered and invoked by MCP-compatible clients at runtime.

Yes, you can organize skill discovery by category by generating a skill-tree.json file, which builds a navigable taxonomy to support structured browsing of skills within multi-skill repositories.

Yes, the skill catalog generator relies on SKILL.md conventions within your skills directory to accurately scan, parse, and index the skill metadata into the required JSON output files.

The best way to automate skill discovery is to generate index.json and embeddings files beforehand, enabling agents to dynamically browse categories and retrieve the right skill at runtime.