The Agent Skills Directory

[COMMAND_EXECUTION]: The Python scripts (velocity_analyzer.py, contributor_insights.py, and code_churn_analyzer.py) use the subprocess.run() function to execute git CLI commands. These operations are required for the skill to mine repository history (e.g., git log, git diff, git rev-parse). The implementation is secure as it passes arguments as a list and does not enable shell execution (shell=False), which prevents shell injection attacks.
[INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes untrusted data originating from a git repository's history.
- Ingestion points: The scripts ingest commit hashes, author emails, commit timestamps, commit subjects, and file-level numstat data from the repository.
- Boundary markers: The skill does not currently implement specific delimiters or 'ignore instructions' warnings when processing commit subjects or other user-generated text.
- Capability inventory: The skill can execute local git commands and write markdown reports to the filesystem using the retro_report_generator.py script.
- Sanitization: Commit data is read and processed as strings without sanitization or escaping before being presented to the agent or included in the final report artifacts. While this is a vulnerability surface, the lack of network access and the restricted scope of the commands make it a low-risk, safe implementation.

sprint-retrospective