workflow-skill-designer
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a meta-utility for designing other skills. It includes a comprehensive validation phase (Phase 4) that checks generated files for prohibited patterns like placeholders (TODO/TBD), raw argument parsing, and undefined functions.
- [COMMAND_EXECUTION]: Uses the
Bashtool for benign filesystem operations such as creating directory structures for new skills (mkdir -p). - [INDIRECT_PROMPT_INJECTION]: While the skill ingests user input and local command files to generate new skill packages, it emphasizes 'Phase File Hygiene' and provides sanitization logic (e.g.,
sanitizePhaseContent) to prevent common prompt injection issues or unstable code from appearing in the generated output. This is documented as a safety measure for the generated skills. - [SAFE]: Implements a
disable-model-invocation: trueflag in its frontmatter, which is a standard safety practice to prevent the skill from being triggered accidentally without explicit intent from the user or the orchestrating agent.
Audit Metadata