The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/audit_semver.py invokes subprocess.run to execute the git tag command for retrieving version information from the repository.
[COMMAND_EXECUTION]: The scripts/test_validators.py file uses subprocess.run to launch Python processes for executing the audit scripts as part of its unit tests.
[INDIRECT_PROMPT_INJECTION]: The skill acts on external project data, which introduces a potential surface for indirect prompt injection if malicious content is embedded in repository files. 1. Ingestion points: Data is ingested from CHANGELOG.md and repository manifests via the audit scripts. 2. Boundary markers: The skill does not employ specific delimiters or instructions to the agent to ignore instructions found within processed data. 3. Capability inventory: Local execution capabilities include git commands and script invocation through the subprocess module. 4. Sanitization: Input is validated against regular expressions for SemVer and changelog formats in scripts/audit_semver.py and scripts/audit_changelog.py.

repository-docs