integrating-agentkit

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the official Scalekit SDK and standard dependencies (e.g., Requests, LangChain, Google ADK) from public registries to enable API connectivity.
  • [PROMPT_INJECTION]: The integration identifies an indirect prompt injection surface as agents process untrusted external data, such as the contents of Gmail messages.
  • Ingestion points: User prompts and Gmail message content processed during API calls and agent invocation.
  • Boundary markers: None defined in the integration examples.
  • Capability inventory: Functionality to read, send, and organize emails via the Gmail API.
  • Sanitization: Not present in the provided basic implementation snippets.
  • [DATA_EXFILTRATION]: The skill facilitates authorized network communication with the Scalekit platform and well-known third-party services (Gmail API) to manage OAuth flows and perform user-requested tasks.
  • [CREDENTIALS_UNSAFE]: The skill explicitly includes guardrails against hardcoding credentials, mandating that sensitive keys (SCALEKIT_CLIENT_SECRET) are managed via environment variables.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 09:52 AM
Security Audit — agent-trust-hub — integrating-agentkit