qwencloud-model-selector

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly allows on-demand fetching of public documentation for "latest" data (references/sources.md links such as https://docs.qwencloud.com/... and https://www.qwencloud.com/models) and the Update Check workflow includes running npx to install/run a sibling skill (npx skills add QwenCloud/qwencloud-ai ... and running its scripts), meaning the agent will fetch and execute content from public third‑party repos/sites which the agent must read/interpret to make model/pricing/update decisions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's Update Check flow explicitly runs a runtime install command that fetches and executes remote code — "npx skills add QwenCloud/qwencloud-ai --skill qwencloud-update-check -y" (repository QwenCloud/qwencloud-ai) and then runs downloaded scripts (e.g., gossamer.py / check_update.py), so external content is fetched at runtime and executed.

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

• Attempt to modify system services in skill instructions detected (medium risk: 0.50). The skill contains mandatory post-execution steps that instruct the agent to run local install/update commands (npx, python scripts) which modify the host environment and install code, so it pushes the agent to change machine state even though it doesn't request sudo or create users.