← 返回 Skills 市场
Smart Model Router
作者
PaleBlueDot AI
· GitHub ↗
· v1.0.6
· MIT-0
171
总下载
0
收藏
0
当前安装
7
版本数
在 OpenClaw 中安装
/install task-based-model-router
功能描述
Optimize Your API Costs - Route tasks to cost-effective models via TokenRouter. Dynamically builds model tiers from live pricing, classifies tasks into 6 cat...
安全使用建议
This skill will read and modify your OpenClaw config (~/.openclaw/openclaw.json), ask you to provide a PaleBlueDot TokenRouter base URL and your real API key, and then store that key and route all models through the provider. The package metadata does not declare these config/credential requirements — treat that as a red flag. Before installing or running: 1) Verify you trust PaleBlueDot and want all model traffic routed through it; 2) Back up ~/.openclaw/openclaw.json; 3) Inspect the skill's bundled scripts (manage_models.py) yourself; 4) Do not paste secrets until you understand where they will be stored; 5) Consider testing in a sandbox or with a limited/throwaway API key and check that you can revert the changes (remove the provider / models.allowed entries). If you want to proceed, ask the author to correct the registry metadata to declare the required config paths and credential so the request is transparent.
功能分析
Type: OpenClaw Skill
Name: task-based-model-router
Version: 1.0.6
The skill bundle is a legitimate model routing and cost optimization tool for the OpenClaw platform. It manages model provider configurations in `~/.openclaw/openclaw.json`, fetches live pricing from `https://open.palebluedot.ai`, and orchestrates multi-agent tasks based on task complexity. While the scripts perform sensitive operations—such as modifying the user's configuration file (`manage_models.py`), reading file contents (`smart_find.py`), and analyzing codebase structure (`smart_map.py`)—these behaviors are strictly aligned with the stated purpose of a model manager and developer utility. The setup flow in `SKILL.md` is transparent, requiring explicit user-provided credentials rather than attempting to steal or hardcode them.
能力评估
Purpose & Capability
The skill's stated purpose (route tasks via a TokenRouter) matches the code and instructions which contact PaleBlueDot's pricing API and update the OpenClaw provider entry. However the registry metadata declares no required env vars or config paths, yet the SKILL.md and scripts explicitly require reading/writing ~/.openclaw/openclaw.json and storing a PaleBlueDot API key. The absent metadata is inconsistent and reduces transparency.
Instruction Scope
SKILL.md explicitly directs the agent to: read ~/.openclaw/openclaw.json, prompt the user for a real PaleBlueDot TokenRouter API key and base URL, run setup/check/sync commands that will write the provided API key into the config, and auto-sync models into models.allowed so all models route through that provider. Reading/writing local config and automatically changing the allowed models list expands scope beyond a passive helper — it changes global agent configuration.
Install Mechanism
There is no external install/download step (instruction-only), and all code is bundled with the skill. The bundled scripts perform network requests to https://www.palebluedot.ai (pricing API) and file operations on ~/.openclaw/*; no remote installers or shortened/personal URLs were found. This is proportionate for the skill's objective but still writes to local config.
Credentials
The registry lists no required credentials, yet SKILL.md requires the user to supply a PaleBlueDot API key and base URL which the agent will store in ~/.openclaw/openclaw.json. The scripts also read/write additional local files (~/.openclaw/workspace/swarm_memory.json, swarm_insights.json, and model-routing.json) that may contain sensitive data. Requesting and persisting a third‑party API key and modifying allow lists is a sensitive capability and should be declared up front.
Persistence & Privilege
Although always:false and autonomous invocation are normal, the skill alters the user's global OpenClaw config: it will add/enable a TokenRouter provider, add all models to models.allowed, and route all models through that provider. This persistent, cross-cutting configuration change increases the blast radius (third‑party routing of all model calls) and should be explicitly consented to and reversible.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install task-based-model-router - 安装完成后，直接呼叫该 Skill 的名称或使用
/task-based-model-router触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.6
Version 1.0.6
- Updated all user-facing copy to refer to "PaleBlueDot AI" instead of "PBD" or simply "TokenRouter" for improved clarity.
- No changes to functionality, features, or logic—documentation and guidance text only.
- Ensured setup and user instructions are consistent with new provider naming.
v1.0.5
task-based-model-router 1.0.5
- Added a dedicated CHANGELOG.md file.
- Overhauled the provider setup flow: The skill now requires real PBD TokenRouter credentials before any usage, with an interactive setup, verification, and model sync phase.
- Provider detection now matches any baseUrl containing "https://open.palebluedot.ai" (not just strict equality).
- Skill performs a check and auto-syncs allowed models before every planning task, ensuring up-to-date configuration.
- SKILL.md guidance updated: clearer instructions for initial setup, config management, and user feedback during onboarding.
- Improved agent behaviors: never writes placeholder API keys, and always confirms with the user after syncing models.
v1.0.4
- Documentation updated to clarify TokenRouter setup steps for new users.
- Instruction for obtaining TokenRouter information reworded for accuracy.
- No changes to logic or core functionality; update is strictly to the SKILL.md documentation.
v1.0.3
No file changes detected for version 1.0.3; documentation and behavior remain unchanged. No new features or bug fixes in this release.
v1.0.2
task-based-model-router 1.0.2 Changelog
- Updated SKILL.md with improved explanations for automatic provider setup and planning features.
- Clarified steps for TokenRouter provider configuration and the planning process.
- Fixed minor documentation issues and typos to enhance readability.
- No code or functional changes; documentation only.
v1.0.1
Version 1.0.1
- Updated documentation in SKILL.md for clarity and accuracy.
- Removed the inline API configuration example for TokenRouter from the setup instructions.
- No changes to core logic or features—this update only modifies the documentation content.
v1.0.0
Smart Model Router v1.0.0
- Initial release of Smart Model Router for OpenClaw.
- Automatically configures and updates TokenRouter provider/models before any planning.
- Dynamically builds model pricing tiers from live TokenRouter API.
- Classifies tasks (English & Chinese) into 6 categories and routes across multi-tier models for max cost savings.
- Supports detailed multi-agent workflows with adaptive fallback and easy model enabling.
- Provides real-time model price listing and intelligent routing recommendations for coding, analysis, writing, creative, translation, and simple tasks.
元数据
常见问题
Smart Model Router 是什么？
Optimize Your API Costs - Route tasks to cost-effective models via TokenRouter. Dynamically builds model tiers from live pricing, classifies tasks into 6 cat... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 171 次。
如何安装 Smart Model Router？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install task-based-model-router」即可一键安装，无需额外配置。
Smart Model Router 是免费的吗？
是的，Smart Model Router 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Smart Model Router 支持哪些平台？
Smart Model Router 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Smart Model Router？
由 PaleBlueDot AI（@yb98k999）开发并维护，当前版本 v1.0.6。
推荐 Skills