mise-expert

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

• Third-party content exposure detected (high risk: 0.80). SKILL.md explicitly instructs fetching and running remote installer content from public third-party sources (e.g., "curl https://mise.run | sh" in the GitLab CI and Docker examples and use of the public GitHub Action jdx/mise-action@v2), so the agent could ingest and act on untrusted external content that could alter tools or behavior.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's CI and Docker examples explicitly run remote installers at runtime (e.g., "curl https://mise.run | sh") and reference external GitHub Action repositories (e.g., uses: jdx/mise-action@v2), which fetch and execute remote code needed to install mise for the workflow to run.