writing-skills

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes untrusted user descriptions to generate new skill files and instructions.\n
  • Ingestion points: User input regarding skill purpose, triggers, and functional logic is collected during the intake phase (SKILL.md Step 0).\n
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used when interpolating user data into the generated SKILL.md template.\n
  • Capability inventory: The skill is capable of writing files to the local file system (SKILL.md, agents/openai.yaml, scripts/) and executing shell commands via node and npx for validation (Step 5).\n
  • Sanitization: User-provided strings are incorporated into the output files without sanitization or structural validation to prevent malicious instruction injection.\n- [EXTERNAL_DOWNLOADS]: Fetches and runs packages from the npm registry using npx (specifically skillcheck and agnix) and installs dependencies via npm install if not already present in the workspace.\n- [COMMAND_EXECUTION]: Executes shell commands including node, npm, and npx to build project components and perform automated validation checks on the skill directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 11:55 AM