skill-creator

Pass

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes bun test test/skills-conformance.test.ts to verify that newly created skill files meet the platform's conformance standards. This is a local shell command execution using the Bun runtime.
  • [INDIRECT_PROMPT_INJECTION]: As a 'skill creator', this tool ingests user descriptions and intents to generate new system instructions (SKILL.md files). This represents an attack surface where malicious user input could influence future agent behavior.
  • Ingestion points: User input defining the purpose and content of the new skill in Phase 1 and Phase 3.
  • Boundary markers: The skill uses a structured YAML and Markdown template to format the generated instructions, providing structural boundaries.
  • Capability inventory: The skill has permissions to search files, list pages, write to the filesystem (mutating: true), and execute local tests via bun.
  • Sanitization: There is no evidence of sanitization or escaping of user-provided strings before they are interpolated into the generated skill's triggers, contract, or phases.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 29, 2026, 06:18 PM
Security Audit — agent-trust-hub — skill-creator