cloudflare

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE]: The scripts/ directory containing deploy-cloudflare.js, which is described as the primary way to use the skill in SKILL.md, is missing from the file set.
  • [CREDENTIALS_UNSAFE]: A hardcoded public key (TEST_PEM) is present in the unit test file worker/src/__tests__/crypto-jwt.test.ts. This is documented as being for testing purposes only.
  • [PROMPT_INJECTION]: The skill provides an AI chat proxy endpoint in worker/src/index.ts that forwards requests to OpenRouter. This presents a potential surface for indirect prompt injection.
  • Ingestion points: The request body is consumed directly from c.req.json() in worker/src/index.ts.
  • Boundary markers: No specific boundary markers or 'ignore' instructions are added to the proxied payload.
  • Capability inventory: The worker can make network requests to openrouter.ai using a configured API key.
  • Sanitization: No content filtering is applied to the incoming data before it is sent to the AI provider.
  • [EXTERNAL_DOWNLOADS]: The worker interacts with the OpenRouter API service to provide AI functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 10:18 AM