@tank/local-llm-fit

Warn

Audited by Snyk on May 31, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

  • Potentially malicious external URL detected (high risk: 0.90). The skill explicitly requires live exact-string lookups of model cards during runtime (e.g. https://huggingface.co/models?search= and https://ollama.com/library/ — plus GitHub/web search patterns shown) and uses the fetched model card data to resolve parameters and drive the agent's fit calculations, so these URLs are runtime dependencies that directly control the agent's prompts/behavior.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

  • Hidden Unicode characters detected (1 type(s) found)

Issues (2)

W012
MEDIUM

Unverifiable external dependency detected (runtime URL that controls agent).

W021
MEDIUM

Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

Audit Metadata
Risk Level
MEDIUM
Analyzed
May 31, 2026, 11:05 PM
Issues
2
Security Audit — snyk — @tank/local-llm-fit