create-skill

Pass

Audited by Gen Agent Trust Hub on May 7, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill's review and upgrade modes process external files, which could contain instructions intended to influence the auditing agent's behavior.\n
  • Ingestion points: Target skill files (SKILL.md, rules/*.md) ingested for analysis in review and upgrade modes.\n
  • Boundary markers: No explicit instructions are provided to use delimiters or ignore instructions within the processed data.\n
  • Capability inventory: The skill can write files and execute local shell commands for symlinking and inventory management.\n
  • Sanitization: New skill names are validated against a strict regex pattern (kebab-case) and checked for reserved words to prevent command injection or path traversal.\n- [DYNAMIC_EXECUTION]: The skill executes shell commands such as ln -s and readlink during the wiring phase to establish directory structures for local development. These operations are restricted to the local filesystem and the project's repository structure.
Audit Metadata
Risk Level
SAFE
Analyzed
May 7, 2026, 07:23 AM