The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The 'Audit agents' workflow requires the agent to read and analyze existing agent definition files (.claude/agents/*.md). This ingestion of external, potentially untrusted content creates a surface for instructions embedded in those files to influence the agent's behavior.
Ingestion points: .claude/agents/*.md (SKILL.md)
Boundary markers: No specific boundary markers or 'ignore' instructions are provided when reading these files.
Capability inventory: The skill uses Read, Write, and Edit tools and executes local Python scripts for validation (SKILL.md, reference/detailed-reference.md).
Sanitization: No sanitization or validation of the content of agent files is documented before they are processed by the agent.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute specific local Python scripts (e.g., validate_skill.py, check_constitutional_compliance.py) to perform automated validation and compliance reporting.

development-management