happyhorse

Fail

Audited by Snyk on Nov 19, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.30). The skill is a documented video-generation wrapper, but its top-line install instruction points to a different publisher (belt-sh/cli) than the inference-sh org used everywhere else, which is a possible lookalike source for an unvetted CLI.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (low risk: 0.10). The required workflow runs belt app run with user-supplied prompts and user-selected image/video URLs, and the outputs are generated media rather than readable outsider text, so the only outsider-influenced input is a user-selected media URL (a low-risk path at most).

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

  • Potentially malicious external URL detected (medium risk: 0.30). The setup line at SKILL.md:15 points to a raw.githubusercontent.com install guide for the required belt CLI, which is an inbound setup source that could supply install instructions the agent follows; the inference-sh organization appears to be the vendor, but the naming mismatch with the belt-sh/cli skill source leaves first-party publisher status unconfirmed, so under recall-first I classify it as a reputable unofficial dependency (medium).

Issues (3)

E004
CRITICAL

Prompt injection detected in skill instructions.

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

W012
MEDIUM

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

Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 19, 2026, 12:06 PM
Issues
3