transcript-downloader

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches untrusted third-party content: scripts/check_update.py downloads the SKILL.md from a raw GitHub URL (and the workflow mandates always running it and re-reading any updated SKILL.md), and scripts/downloader.py (via mr-transcript) fetches public YouTube transcripts — both of which are user/third-party content that the agent is required to read and which can change subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.90). scripts/check_update.py performs a runtime fetch of the SKILL.md from https://raw.githubusercontent.com/{owner}/{repo}/main/.agents/skills/{skill_name}/SKILL.md, and the fetched metadata is used to produce an UPDATE_COMMAND (e.g., "npx skills update ...") that the skill's workflow instructs the agent to run, meaning remote content can directly control agent instructions and trigger execution of remotely fetched code.