alquimia-sdk

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for building applications that process untrusted data (chat text, attachments, and custom metadata) and pass it to an AI backend. \n
  • Ingestion points: AssistantInput in SKILL.md, addAttachments in features/attachments.md, and the ExtraData tool in features/tools.md. \n
  • Boundary markers: The documentation does not demonstrate the use of explicit delimiters or instructions to prevent the model from misinterpreting untrusted data. \n
  • Capability inventory: The SDK is capable of making network calls to a configured inference backend and third-party services like ElevenLabs. \n
  • Sanitization: No evidence of input sanitization or filtering was found in the provided integration examples.\n- [EXTERNAL_DOWNLOADS]: The skill references several Node.js packages to be installed via npm. These include vendor-specific packages (@alquimia-ai/tools, @alquimia-ai/ui) and reputable open-source libraries (hono, tailwindcss, framer-motion, lucide-react). All sources are considered safe or vendor-managed.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 04:52 PM
Security Audit — agent-trust-hub — alquimia-sdk