The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (e.g., python .claude/scripts/code_graph) to analyze repository structure and trace connections during the context discovery phase.
[PROMPT_INJECTION]: The skill demonstrates a surface for indirect prompt injection by ingesting and processing content from various repository files which may be untrusted. 1. Ingestion points: Reads project documentation and configuration files including CLAUDE.md, docs/project-config.json, and files within the docs/project-reference/ directory. 2. Boundary markers: No explicit delimiters or instructions are provided to the sub-agent advisors to ignore or treat embedded instructions as data within the analyzed files. 3. Capability inventory: The workflow involves spawning multiple sub-agents (spawn_agent), executing local shell commands, and writing files to the repository. 4. Sanitization: There is no evidence of sanitization, filtering, or escaping of the content read from project files before it is passed to the advisors or the chairman.

llm-council