spine-animation

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md’s runtime pipeline ingests user-supplied free text in split_character.py by sending the user-provided character image to Google Gemini with fixed POSITIVE_PROMPT/NEGATIVE_PROMPT and requesting IMAGE+TEXT generation (no user-message selection gate for the LLM input content).

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

• Potentially malicious external URL detected (high risk: 0.80). The skill-generated preview HTML loads and executes the official Spine Web Player from the UNPKG CDN at runtime (https://unpkg.com/@esotericsoftware/spine-player@4.2.*/dist/iife/spine-player.js and its CSS), which is a required runtime dependency that pulls and runs remote JS in the user's browser.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (medium risk: 0.40). The skill instructs the agent to write multiple scripts to /home/claude and to run pip install with the --break-system-packages flag (which bypasses packaging safeguards and can alter the host Python environment), so it encourages modifying the host environment even though it does not request sudo, create users, or edit privileged system files.