kelet-integration

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute curl requests against the Kelet API (api.kelet.ai) for configuring synthetic evaluators. It also uses a python3 one-liner to generate base64-encoded deeplinks for manual configuration fallback.
  • [EXTERNAL_DOWNLOADS]: The skill manages the installation of the kelet library and essential OpenTelemetry peer dependencies (@opentelemetry/api, @opentelemetry/sdk-trace-node, etc.) from official public registries (PyPI and npm).
  • [DATA_EXFILTRATION]: Transmits project identifiers and evaluator definitions to the vendor's infrastructure at api.kelet.ai. This data transfer is a core requirement for the stated purpose of the skill to enable remote RCA and monitoring.
  • [CREDENTIALS_UNSAFE]: Requests the user's KELET_API_KEY to perform administrative setup tasks. The skill provides extensive guidance on securely storing these credentials in production environments using platform-specific secrets managers (e.g., Vercel, Fly.io, Kubernetes Secrets) and specifically warns against committing them to git history.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 08:49 AM