turtle-harden

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to perform shell commands in the 'SEAL' (verification) phase, specifically pnpm install, gw ci, and grep. These commands are used to synchronize project dependencies, run continuous integration checks, and scan the codebase for hardcoded secrets or dangerous patterns like eval. These actions are consistent with the skill's purpose as a security audit and hardening tool.
  • [INDIRECT_PROMPT_INJECTION]: Because the skill is designed to ingest and analyze potentially untrusted code from users, it possesses an attack surface for indirect prompt injection. However, the skill explicitly provides the agent with instructions and code examples for implementing robust sanitization, validation, and boundary markers (e.g., using Zod schemas and DOMPurify) to mitigate these risks when the agent generates or modifies code.
  • [SAFE]: The skill serves as a security best-practice guide and does not contain any malicious patterns. It does not attempt to exfiltrate data, maintain persistence, or obfuscate its instructions. The mention of sensitive patterns (e.g., 'sk-live', 'AKIA') is restricted to diagnostic grep commands intended to help the agent find and report potential security flaws in the user's code.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 01:31 AM