close
Warn
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill logic identifies a script location at runtime by searching for directories under
~/.claude/plugins/cache/and proceeds to execute a file namedclose-context.shfrom the discovered path. - [COMMAND_EXECUTION]: The skill performs several shell-based operations including directory traversal with
findandls, git repository management (staging and committing changes), and piping tool output into inline Python scripts for JSON processing. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external tool outputs (specifically
bon list --json) to generate action items and proposals. While the processing is handled via structured JSON parsing in Python, this creates an ingestion surface where untrusted data could influence agent recommendations during the session reflection phase.
Audit Metadata