ratel-tune-definitions
Warn
Audited by Snyk on Jul 14, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.85). Yes—Step 2 fetches up-to-date Ratel docs from public web sources (e.g.,
https://docs.ratel.sh/...andhttps://raw.githubusercontent.com/...) and the workflow then uses that fetched free text as LLM context for the tuning plan.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs runtime fetching of Ratel docs (e.g., https://docs.ratel.sh/llms.txt, https://docs.ratel.sh/llms-full.txt and https://raw.githubusercontent.com/ratel-ai/ratel/main/README.md) and tells the agent to "trust the docs" to determine BM25/tokenization and the Skill data model, so the fetched content will directly control the agent's prompt/instruction behavior at runtime.
Issues (2)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W012
MEDIUMUnverifiable external dependency detected (runtime URL that controls agent).
Audit Metadata