medeo-video

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs the agent to ask users to paste their Medeo API key (starts with "mk_") and then run a command that embeds that key verbatim as a CLI argument (config-init --api-key "mk_..."), which requires the LLM to handle/output the secret directly (high exfiltration risk).

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly downloads and ingests arbitrary public/user-generated media (e.g., SKILL.md sections 3a/3b, docs/assets-upload.md and scripts/medeo_video.py: cmd_upload, cmd_upload_file, upload_media_urls) — including Discord CDN links and generic HTTP(S) URLs — and those untrusted assets are incorporated into the video-generation workflow (affecting settings, asset_sources, and the AI backend's outputs), so third-party content can materially influence subsequent tool actions.