raison-sdk
Warn
Audited by Snyk on Feb 22, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 1.00). The SDK fetches prompt templates from the external raison.ist service (real-time WebSocket sync / Raison.render as described in SKILL.md and references/sdk-reference.md) and the examples explicitly pass those rendered prompts as system messages to LLMs (references/examples.md), so third-party prompt content can directly influence agent behavior.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.90). The Raison SDK opens a runtime WebSocket to https://api.raison.ist (prompts are hosted at raison.ist) and directly fetches prompt templates that are used as system prompts for LLM calls, so this external URL controls agent instructions at runtime.
Audit Metadata