The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/pr_writer.py uses subprocess.run to execute local git commands for retrieving branch information, commit history, and file diffs. These calls use argument lists rather than shell strings, which mitigates standard shell injection risks.\n- [DYNAMIC_EXECUTION]: The test script scripts/test_cli.py uses the importlib library to dynamically load and execute modules for testing purposes. This is restricted to local scripts within the skill's expected directory structure.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local repository, including git commit messages and project configuration files (package.json, pubspec.yaml, go.mod), to populate the PR description template.\n
Ingestion points: Data is pulled from git logs via commits() and changed_files() in scripts/pr_writer.py, and from project manifest files via detect_framework().\n
Boundary markers: None identified in the generated markdown output.\n
Capability inventory: The skill has the capability to execute shell commands via subprocess.run in scripts/pr_writer.py and scripts/test_cli.py.\n
Sanitization: Content is processed into markdown strings but no specific escaping or sanitization of commit message content is performed prior to inclusion in the draft.

pr-writing