The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and processing untrusted external data.
Ingestion points: Fetches Dependabot alerts via gh api, release notes via gh release view, and downloads/extracts changelogs from PyPI source distributions (sdist tarballs) in SKILL.md.
Boundary markers: The instructions do not define explicit delimiters or instructions for the agent to ignore potentially malicious content within the fetched data.
Capability inventory: The skill possesses significant capabilities, including shell command execution (git, gh, python, pip, poetry, uv, etc.), file system modification (manifests and lockfiles), and network access.
Sanitization: No escaping, validation, or filtering of the external content is performed before processing.
[COMMAND_EXECUTION]: The skill frequently executes shell commands to interact with the environment and process data.
Python processing: Uses python -c and heredoc scripts (`python
<<EOF) in SKILL.mdto dynamically parse TOML files (likepoetry.lock, uv.lock, pdm.lock), compare versions using the packaging` library, and normalize package names. These are implemented as static templates for data processing.
System tools: Orchestrates git, gh, ripgrep (rg), and find for repository analysis and triaging tasks.
[EXTERNAL_DOWNLOADS]: The skill fetches data and metadata from well-known and trusted external services.
GitHub: Retrieves vulnerability alerts and release notes using the gh CLI (api.github.com).
PyPI: Fetches package metadata and source distributions from pypi.org using curl to determine patched versions and extract changelogs.

dependabot-triage-py