The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a standard git log command pipeline (git log --all --diff-filter=M --name-only --pretty=format: | sort | uniq -c | sort -rn | head -20) to identify frequently modified files. This is a read-only operation intended for churn analysis and does not pose a security risk in this context.
[DATA_EXFILTRATION]: The skill accesses repository documentation (e.g., CLAUDE.md, README, contributing guides) and checks for the existence of specific file types (images, fonts, audio). This data exposure is localized to the agent's context for the purpose of workflow discovery and does not involve external network transmission.
[PROMPT_INJECTION]: The skill includes an assessment of untrusted repository content (Indirect Prompt Injection surface). It reads docs and scripts to find manual steps. While this is an ingestion point for external data, the skill's logic is focused on structural analysis and prescribing human-implemented fixes rather than executing instructions found within the analyzed files. This is considered a low-risk, functional requirement of the skill.

loop-check