The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/check_supply_chain.py uses subprocess.run to invoke the gh CLI. This functionality is used to programmatically query repository metadata, security status, and vulnerability alert configurations directly from GitHub.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to download and utilize security tools from official package registries and well-known services.
It utilizes uvx to fetch and run pip-audit and cyclonedx-bom from PyPI.
It uses the gh CLI to interact with the GitHub API for repository configuration, SBOM generation, and security auditing.
[REMOTE_CODE_EXECUTION]: The skill uses uvx to download and execute pip-audit and cyclonedx-py directly from PyPI. These tools are executed with pinned versions (==2.7.3 and ==4.4.3 respectively) to mitigate the risk of adopting recently published malicious releases.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from a local repository and possesses significant execution capabilities.
Ingestion points: The scripts/check_supply_chain.py script reads multiple configuration files from the target repository, including .github/dependabot.yml, lockfiles (e.g., uv.lock, poetry.lock), CODEOWNERS, and GitHub Actions workflow files.
Boundary markers: The skill does not implement explicit delimiters or "ignore embedded instructions" warnings when the agent processes the content of these external files.
Capability inventory: The skill can execute shell commands via subprocess.run, perform authenticated GitHub API operations via the gh CLI, and run Python tools via uvx.
Sanitization: The audit script employs conservative regex-based matching for parsing YAML and TOML files to minimize risks associated with complex parser exploits.

python-supply-chain