The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface for indirect prompt injection due to its handling of external data.
Ingestion points: Examples in SKILL.md and references/async-patterns.md describe reading files via pathlib and fetching URL content via httpx.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions embedded within the files or web content it reads.
Capability inventory: The skill utilizes file system access (pathlib), network operations (httpx), and the execution of external tools (mypy, black, ruff, pytest).
Sanitization: No input validation or sanitization logic is shown in the provided code templates.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of various Python packages and the use of external repositories for development workflow automation.
Packages: References standard libraries and tools such as mypy, black, ruff, pytest, httpx, and pydantic in project configuration examples.
Remote sources: Fetches pre-commit hook configurations from the official repositories of Black (github.com/psf/black) and Ruff (github.com/astral-sh/ruff-pre-commit).
[DATA_EXFILTRATION]: The skill includes code patterns that involve network access and local file system path resolution.
Network operations: The httpx.AsyncClient is used in examples to fetch data from lists of URLs.
Sensitive file paths: The references/standard-library.md file contains an example that resolves the user's home directory using Path.home().

python-pro