The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the local command resume-builder verify to perform deterministic provenance checks on resume files. This is a functional requirement for the skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted resume data, which theoretically presents an indirect prompt injection surface. However, the skill's primary function is to audit this data for errors. Ingestion points: resume.yaml (argument path) and corrections.yaml (local project file). Boundary markers: No explicit delimiters are present to separate resume content from potential malicious instructions. Capability inventory: The skill can execute the resume-builder command and append data to session logs. Sanitization: No explicit sanitization or filtering of the input file content is performed.

verify