The Agent Skills Directory

[COMMAND_EXECUTION]: The Python scripts scripts/agent.py and scripts/process.py use subprocess.run to call the cosign CLI tool.
Arguments are passed as a list rather than a single string, which prevents shell-based command injection.
The execution is constrained to the cosign binary and is necessary for the skill's primary functionality of signing and verifying images.
[EXTERNAL_DOWNLOADS]: The SKILL.md file contains instructions for downloading the cosign binary directly from its official GitHub releases repository (sigstore/cosign).
This source is a well-known and trusted repository for security tooling.
[COMMAND_EXECUTION]: Installation steps in SKILL.md include shell commands requiring administrative privileges (sudo mv, sudo chmod) to install the binary to system paths.
These instructions are for manual user execution and represent standard installation procedures for system-level CLI tools.
[COMMAND_EXECUTION]: The script scripts/process.py processes image lists provided via external files, which are then used in command execution.
Ingestion points: Image references are read from a file via Path(args.images_file).read_text() in scripts/process.py.
Boundary markers: No explicit boundary markers or 'ignore' instructions are used for the input data.
Capability inventory: The script uses subprocess.run to execute commands and Path.write_text to generate reports.
Sanitization: No explicit sanitization is performed on image strings before they are passed as arguments; however, the use of argument lists in subprocess.run prevents traditional shell injection.

implementing-image-provenance-verification-with-cosign