ai-orchestration-vercel-ai-sdk
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: Hardcoded dummy credential pattern found in educational examples. While explicitly marked as a 'Bad Example' to demonstrate insecure practices, the presence of the 'sk-' prefix matches credential detection patterns. Evidence found in
examples/core.md(apiKey: "sk-proj-abc123def456"). - [INDIRECT_PROMPT_INJECTION]: The skill facilitates Retrieval-Augmented Generation (RAG) workflows where untrusted data is processed. This creates a surface for indirect prompt injection if retrieved documents contain malicious instructions.
- Ingestion points: Untrusted document content is ingested via the
ragQueryandstreamRagQueryfunctions inexamples/rag.md. - Boundary markers: The implementation uses text delimiters like 'Context:' and 'Question:' to separate data from instructions, which provides basic but incomplete protection.
- Capability inventory: The skill enables
generateText,streamText, and tool calling capabilities which could be manipulated by injected instructions. - Sanitization: No explicit sanitization or filtering of retrieved document content is shown before it is interpolated into the model prompt.
Audit Metadata