charles-color-analysis

Fail

Audited by Snyk on Nov 16, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.30). The skill's colour-analysis purpose is legitimate and its scripts make no network calls or covert actions, but dom_colors.js reads live form values (el.value), including password fields, into a sample string that gets saved to disk, which is a careless exposure of user data rather than malicious intent.

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

  • Third-party content exposure detected (low risk: 0.10). The workflow reads text from user-supplied images (macOS Vision OCR in measure) and from user-selected web pages (dom_colors.js walks all page text), which are outsider-authored content the agent reaches only after the user picks a specific item, so this is low risk.

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

  • Potentially malicious external URL detected (medium risk: 0.30). The README's npx skills add LemonFace0309/charles-skills is an instructed setup path that installs skill material from an individual-owned repository, which is unofficial but shows no warning signs, so it is a common/reputable unofficial dependency and rates medium; the scripts and references contain no other remote locators.

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 16, 2026, 05:26 PM
Issues
3