cost-benchmark

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute a local JavaScript file (bench.mjs) using Node.js. This is a standard operation for running performance benchmarks within a development repository.
  • [CREDENTIALS_UNSAFE]: The instructions describe the retrieval of sensitive API keys (GOOGLE_AI_API_KEY, ANTHROPIC_API_KEY) from gcloud secrets or via environment variable overrides. While this involves sensitive data, it follows standard secret management practices for developer tools.
  • [PROMPT_INJECTION]: The skill ingests data from a 'structural+adversarial corpus' (booster-corpus.json). This represents an indirect prompt injection attack surface where untrusted data is processed by the agent, though the stated purpose is for diagnostic and benchmarking purposes.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 01:22 PM
Security Audit — agent-trust-hub — cost-benchmark