The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/collect-source-evidence.py uses subprocess.run to execute local git commands (e.g., git log, git rev-parse) to gather project history. The implementation uses list-based arguments rather than string-based shell execution, which follows best practices to prevent command injection.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting and processing external data from project repositories (markdown files and commit messages). This creates an inherent surface for indirect prompt injection if the source files contain adversarial instructions. However, the skill provides specific instructions in SKILL.md and references/distillation-lens.md to filter private data and verify results through human review, mitigating the risk.
[SAFE_PRACTICE]: The instructions explicitly guide the user and agent to exclude sensitive information such as names, credentials, and private paths during the distillation process, demonstrating a security-conscious design.

sgc-skill-distiller