qwencloud-model-selector

[QwenCloud] Recommend the best Qwen model and parameters. TRIGGER when: choosing between Qwen models, comparing Qwen model pricing, understanding Qwen model...

This skill is instruction-only and appears to do what it says: recommend Qwen models and parameters. It does not ask for keys or install software. Two practical cautions: (1) The agent-compatibility instructions include scanning certain skill directories and appending a registration block to agent config files (e.g., CLAUDE.md, ~/.claude/CLAUDE.md). Make sure the agent asks you before it reads or writes those files and review the exact changes it proposes. (2) The included model/pricing data is a point-in-time snapshot — if you need up-to-date prices or availability, explicitly request the skill to fetch or check the official pages or use a dedicated usage/billing skill. If you prefer stricter control, deny any file-modification requests or run the skill in a sandboxed agent that cannot write your config files.

Type: OpenClaw Skill Name: qwencloud-model-selector Version: 0.2.0 The skill bundle contains instructions in SKILL.md and agent-compatibility.md that direct the AI agent to perform high-risk operations, including modifying local configuration files (CLAUDE.md/AGENTS.md) and executing shell commands. Specifically, the 'Update Check' section mandates the agent to prompt users to install external packages via 'npx' and execute a local script named 'gossamer.py', the source code for which is not provided in the bundle. While these actions are framed as maintenance features and include instructions to ask for user consent, the lack of transparency for the executed scripts and the push for additional package installation represent a significant security risk and potential supply-chain vector.