The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/check_release_deployments.sh uses python -c to parse manifest files by interpolating the file path directly into a Python string literal. This pattern is vulnerable to arbitrary code execution if a file path in the repository contains single quotes or other Python-executable characters.
[COMMAND_EXECUTION]: The skill depends on the just task runner and several specific recipes (normalize-release-tag, release-tag-to-py-version, python-package-paths) which are external to the skill's source code. These commands perform opaque operations on the repository that are not verifiable within the skill's context.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to malicious repository content. An attacker who can influence the repository's file structure could create filenames designed to exploit the unsafe Python interpolation, leading to arbitrary code execution when the checker is run.
Ingestion points: File paths discovered via find and git ls-files in scripts/check_release_deployments.sh.
Boundary markers: None present; the skill assumes all discovered file paths are legitimate.
Capability inventory: Includes python -c for arbitrary Python execution, bash for shell commands, and gh for GitHub API interactions.
Sanitization: The script lacks sanitization or validation of manifest file paths before they are used to construct and execute Python and shell commands.

check-release-deployments