Smart Model Switcher

智能模型自动切换 V4 - 多模态感知，自动识别图片/代码/文本任务，切换到最适合的模型。支持图片理解(qwen3-vl-plus)、代码(glm-5/qwen-coder)、长文本、推理等场景。零感知切换，无需手动操作。

This skill is coherent with its stated purpose (auto-detect task type and switch models). However: 1) The skill instructs agents to run detection before every response and to use session_status or a local gateway API — confirm your gateway's /api/session/model endpoint requires proper auth and that sessionKey values cannot be abused to switch other users' sessions. 2) The SKILL.md contains a detected 'unicode-control-chars' pattern (possible prompt-injection); review the file for hidden characters or injected directives. 3) Test in a sandbox first: verify switching behavior only affects intended sessions, that the local endpoint doesn't accept unauthenticated requests, and that provider API keys remain under your control. 4) If you do not trust the author or cannot verify gateway protections, do not install in a production environment. If you want higher assurance, request the author to remove hidden/control characters and to document authentication/authorization requirements for the gateway API and sessionKey usage.

Type: OpenClaw Skill Name: smart-model-switcher-v4 Version: 4.0.0 The skill bundle is designed to automate model switching within the OpenClaw environment based on user input (e.g., switching to vision models for images or specialized coding models). The logic is implemented through a local Node.js script (scripts/auto-switch.js) that communicates with the OpenClaw gateway via a local API (localhost:3737) and provides clear instructions in SKILL.md for the AI agent to manage session states. No evidence of data exfiltration, malicious intent, or unauthorized remote execution was found.