ask-ai
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill configures a RAG-powered AI chat ("Ask AI") which retrieves documentation content to generate answers. This represents a potential surface for indirect prompt injection if the documentation contains malicious instructions.\n
- Ingestion points: Documentation content retrieved from files/sections via configured search providers (Algolia, Typesense, or MCP search_docs tool).\n
- Boundary markers: The configuration guide does not mention the use of delimiters or instructions to ignore embedded prompts in the retrieved context.\n
- Capability inventory: The feature performs LLM generation based on retrieved context and can infer package names or import statements from the documentation.\n
- Sanitization: No explicit sanitization or validation mechanisms for retrieved documentation are described in the configuration options.
Audit Metadata