The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to run included scripts list-skills.sh and validate-skill.sh. These scripts are used for scanning skill directories, extracting metadata, and verifying that the structure of a skill (frontmatter, required sections) meets workspace standards.
[INDIRECT_PROMPT_INJECTION]: Because the skill processes free-form user prompts to generate or improve other instructions (skills), it possesses an indirect prompt injection surface. The skill manages this risk by requiring a mandatory PLAN block for user transparency before execution, using isolated specialist agents with narrowly defined roles, and providing a review-prompt.md checklist that explicitly instructs the agent to check for hardcoded secrets and command injection in generated code.
[SAFE]: No malicious patterns, such as unauthorized data exfiltration, hardcoded credentials, or persistence mechanisms, were detected. The skill's activities are scoped to specific development directories and follow established platform best practices for agent autonomy and oversight.

skills