The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local validation scripts (pnpm validate:agent-skills, pnpm validate:skill-routing). These commands are used to verify the integrity of the repository's skill structure, frontmatter, and routing logic.
[INDIRECT_PROMPT_INJECTION]: The skill processes and reorganizes other agent skills within the repository, which creates an attack surface for indirect prompt injection if those skills contain malicious instructions.
Ingestion points: Reads SKILL.md files and associated reference documents within the .agents/skills directory.
Boundary markers: The skill format mandates YAML frontmatter to delimit metadata, and validation scripts are used to enforce this structure.
Capability inventory: The skill allows creating, updating, and reorganizing files within the repository, as well as executing local shell commands for validation purposes.
Sanitization: Relies on deterministic validation scripts to check for known command names and stale guidance references during the maintenance process.

skill-maintenance