The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documents the installation of the uv package manager using a piped shell script from the well-known source https://astral.sh/uv/install.sh.
[COMMAND_EXECUTION]: Instructions in references/file-processing.md include using sudo apt-get install for system dependencies such as tesseract-ocr, poppler-utils, pandoc, and libreoffice.
[REMOTE_CODE_EXECUTION]: The documentation mentions using npx to execute the @modelcontextprotocol/inspector tool for testing MCP servers.
[PROMPT_INJECTION]: The patterns provided for reading and extracting text from PDF, Word, and Excel files in references/file-processing.md present an indirect prompt injection surface when the output is processed by an LLM.
Ingestion points: File processing utilities in references/file-processing.md that read external .pdf, .docx, and .xlsx files.
Boundary markers: No explicit instruction delimiters or boundary markers are used in the provided text extraction snippets.
Capability inventory: The skill includes patterns for database operations (references/database-orm.md), file system modifications, and network requests (references/mcp-servers.md).
Sanitization: While the skill includes a safe_path utility in references/mcp-servers.md to prevent directory traversal, it does not provide general sanitization patterns for LLM-directed text processing.

python-dev