scaffold-exercises
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes standard system commands including
mkdir -p,git mv,git commit, andpnpm ai-hero-cli internal lintto scaffold and validate exercise structures. These operations are consistent with the skill's stated purpose. - [PROMPT_INJECTION]: The skill processes an external 'plan' to determine the names and types of directories and files to create, which constitutes an indirect prompt injection surface.
- Ingestion points: Workflow step 1 ('Parse the plan') ingests untrusted plan data.
- Boundary markers: Absent; there are no instructions to ignore embedded commands or specific delimiters for the plan content.
- Capability inventory: Includes directory creation (
mkdir), version control operations (git), and local package execution (pnpm). - Sanitization: Absent; the instructions do not describe any validation or escaping of the plan data before it is used in shell commands.
Audit Metadata