The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git log --oneline -10 to provide experts with a summary of recent project changes. This is a legitimate use of command execution to gather context.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing project files as data for sub-agents. 1. Ingestion points: CLAUDE.md, README.md, .claude/rules/, and the output of git commands. 2. Boundary markers: Ingested data is placed under a 'Data:' heading in the sub-agent prompts. 3. Capability inventory: The skill can trigger the Task tool for sub-agent execution and write report files to the docs/ folder. 4. Sanitization: The skill does not perform sanitization on ingested file content before inclusion in prompts. This is a low-risk architectural pattern for analysis tools.

ceo-council