The Agent Skills Directory

[COMMAND_EXECUTION]: The skill leverages standard command-line utilities (ruff, black, mypy, pytest) and Django management commands to validate code quality and application state. These commands are typical for development and CI/CD workflows.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes local project files and git diffs to detect issues. While reading untrusted project data is an injection surface, the risk is mitigated by the use of structured verification tools and the intended use-case of developer-initiated verification.
Ingestion points: Project source code, requirements.txt, and git diff output (SKILL.md).
Boundary markers: None explicitly defined in the provided commands, though tools like bandit and ruff have their own parsing logic.
Capability inventory: Shell command execution, file system access, and database interactions via Django's management interface.
Sanitization: Standard verification tools provide their own parsing and validation of input files.
[DYNAMIC_EXECUTION]: The skill uses Python heredocs and one-liners to perform complex configuration checks. These are used for introspection of the local environment and are not used to execute remote or untrusted code strings.
[EXTERNAL_DOWNLOADS]: The skill configuration references official GitHub Actions and standard Python packages (bandit, safety, pip-audit) from well-known registries. These dependencies are consistent with the skill's stated purpose.

django-verification