The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install the edgartools library and its AI extensions from the official Python Package Index (PyPI) using pip or uv. It also references the installation of an MCP server via uvx.
[PERSISTENCE]: The skill includes an install_skill() function that persists the agent's capabilities by installing files to the ~/.claude/skills/edgartools/ directory. This is the documented and intended method for extending the agent's functionality with this library.
[INDIRECT_PROMPT_INJECTION]: The skill represents a potential surface for indirect prompt injection as it is designed to ingest and process external content from SEC EDGAR filings, which could contain maliciously crafted text intended to influence agent behavior.
Ingestion points: Reads company filings (10-K, 10-Q, 8-K) and structured XBRL data from the SEC EDGAR database.
Boundary markers: The documentation suggests converting filings to Markdown for LLM processing but does not explicitly demonstrate the use of strict delimiters or injection warnings in the provided code snippets.
Capability inventory: The skill can perform network operations to fetch filings and write processed data to local files (e.g., filing.save() and filing.markdown()).
Sanitization: No explicit sanitization or filtering logic is demonstrated in the provided reference materials for processing raw filing text.

edgartools