error-debugger

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE] (HIGH): The skill documentation (Example 4) demonstrates the agent accessing sensitive user credentials through git config and GitHub authentication tokens via gh auth status.
  • [COMMAND_EXECUTION] (HIGH): The agent is shown executing a wide range of shell commands (npm, npx, git, gh, ping, curl) to perform its tasks, which is a major security risk if the agent's logic is subverted.
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill performs automatic installation of Node.js packages (npm install @prisma/client) based on its own reasoning, which can be leveraged to install malicious packages via dependency confusion or typosquatting.
  • [DATA_EXFILTRATION] (MEDIUM): The skill demonstrates the ability to use curl and fetch for network requests. This capability, combined with its access to auth tokens, could be used to exfiltrate sensitive data.
  • [PROMPT_INJECTION] (LOW): The debugging workflow (Example 3) involves processing raw, untrusted data such as server logs, creating a vulnerability to Indirect Prompt Injection without documented sanitization measures.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 06:06 PM