langchain
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by interpolating untrusted data into prompt templates without boundary markers or validation mechanisms.\n
- Ingestion points: The skill ingests untrusted data through user-provided parameters like 'input', 'message', and 'question', as well as external content retrieved from document stores in the RAG chain implementation.\n
- Boundary markers: Prompts such as 'Answer using context: {context}' do not employ delimiters (e.g., XML tags, triple backticks) or specific instructions to ignore potentially malicious commands embedded within the external context.\n
- Capability inventory: The skill metadata configuration requests access to the 'Bash' and 'Write' tools, and the Python code examples demonstrate the ability to perform network requests using the 'httpx' library.\n
- Sanitization: There is no evidence of content filtering, escaping, or schema validation for the external data before it is interpolated into the agent's context.
Audit Metadata