The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to verify the conformance of newly created skills. Specifically, it calls bun test test/skills-conformance.test.ts during the verification phase and mentions gbrain check-resolvable in the error handling section.
[INDIRECT_PROMPT_INJECTION]: The skill generates new SKILL.md files by interpolating user-provided intent, names, and trigger phrases into a template. This process creates a vulnerability surface where a user could provide malicious instructions that get embedded into the generated skill's contract or phases.
Ingestion points: User-supplied descriptions, triggers, and capability gaps are used to populate the new skill's SKILL.md file in Phase 3.
Boundary markers: The skill does not define specific delimiters or instructions to sanitize user input before it is written to the new skill file.
Capability inventory: The skill has permissions to write to the file system (updating manifest.json, RESOLVER.md, and creating new directories) and execute shell commands (bun test).
Sanitization: No sanitization or validation steps for user-provided text are specified in the generation workflow.

skill-creator