The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads the graphifyy package from PyPI via pip or uv. It also clones remote repositories from GitHub and fetches content from external URLs (YouTube, arXiv, etc.) for ingestion into the knowledge graph.
[COMMAND_EXECUTION]: The skill uses shell commands and Python scripts to perform file detection, AST extraction from code, clustering, and data exports. It includes logic to locate the correct Python interpreter by reading binary shebangs and executing diagnostic Python blocks.
[PERSISTENCE]: The skill includes functionality to install a Git post-commit hook (graphify hook install), ensuring that the knowledge graph is automatically rebuilt whenever new code is committed to a repository.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted files (code, documents, images) to build a knowledge graph, creating a potential surface for indirect prompt injection. Ingestion points include any file within the provided INPUT_PATH, cloned repositories, or URLs fetched via graphify add. The extraction process relies on LLM subagents without explicit boundary markers for untrusted content.
[DYNAMIC_EXECUTION]: The skill can instantiate an MCP (Model Context Protocol) server to expose graph query tools to other agents and supports multi-agent coordination for parallel semantic extraction using platform-specific spawning mechanisms.

graphify