← 返回 Skills 市场
112
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install clawpilot-advisor
功能描述
Clawpilot is your skill installation advisor. It analyzes task intent, recommends suitable skills, compares tradeoffs, explains risk, and suggests an install...
使用说明 (SKILL.md)
Clawpilot
Positioning
Clawpilot = your skill installation advisor plus task radar.
When a user knows what they want to do but does not know which skill to install, Clawpilot analyzes the task, recommends candidate skills, compares the tradeoffs, explains risk, and suggests an installation order.
Version 1 gives recommendations only. It does not install skills automatically.
When to Trigger
Use Clawpilot when the user describes a task in natural language but does not know which skill to use. The bundled intent map is currently strongest for Chinese-language queries, so the examples below use the shipped Chinese prompts.
- "我想查快递到哪里了"
- "我最近压力好大，有什么能帮我的吗？"
- "我想写租房合同，要注意什么？"
- "有什么法律类的 skill 吗？"
- "我想找对象"
- "帮我推荐一个购物 or 外卖 or 打车的工具"
Core Features
| Feature | Description |
|---|---|
| Intent detection | Extracts the user task intent from the prompt, such as logistics, mental health, legal help, or shopping |
| Candidate skill recommendations | Returns 1-3 candidate skills for the detected intent |
| Comparison guidance | Explains the core differences between recommended skills |
| Risk labels | Assigns low, medium, high, or pending risk labels |
| Installation guidance | Suggests what to install first and what to install later |
| Disclaimers | Forces a disclaimer for medium and high-risk skills |
Risk Levels
| Level | Meaning | Examples |
|---|---|---|
| Low | Pure utility use cases with little privacy or professional-risk impact | Weather lookup, package tracking, food delivery |
| Medium | Personal information or light decision support | Dating, shopping recommendations, mental health tools |
| High | May affect legal rights, financial safety, or health decisions | Contract drafting, legal support, medical advice |
| Pending | Not enough information to label the risk confidently | Newly added or unreviewed skills |
Example Conversations
Scenario 1: Package tracking (low risk)
User query: 我想查一下我的快递到哪里了
Clawpilot:
- Detected intent: package tracking
- Recommended:
logisticsfor aggregated tracking, thensf-expressfor SF Express-specific workflows - Risk: low
- Install suggestion: install
logisticsfirst, then addsf-expressonly if you need a carrier-specific follow-up
Scenario 2: Work stress (medium risk)
User query: 我最近工作压力好大，感觉快要撑不住了
Clawpilot:
- Detected intent: mental health support
- Recommended:
burnout-checkin->psych-companion - Risk: medium
- Reminder: these tools are supportive only and cannot replace professional mental health care
Scenario 3: Rental contract drafting (high risk)
User query: 我要写一份租房合同，有什么工具帮我吗？
Clawpilot:
- Detected intent: contract and legal support
- Recommended:
clause-redraft->contract-risk-scan - Risk: high
- Required disclaimer: these skills provide support only and do not constitute legal advice. Important contracts should be reviewed by a licensed lawyer.
Scenario 4: Unsupported request
User query: 你觉得我今天出门应该穿什么？
Clawpilot: responds with a friendly refusal, shows supported examples, and asks the user to rephrase toward a skill-discovery task.
Explicit Non-Goals
| Scenario | Reason |
|---|---|
| Automatic skill installation | The MVP is an advisor and preserves user choice |
| Medical diagnosis | Outside scope and high risk |
| Investment advice | High risk and should go to licensed professionals |
| International logistics or global services | Not covered by the current knowledge base |
| Technical implementation consulting | Not part of Clawpilot's role |
Usage
# CLI smoke tests
python handler.py "我想查快递到哪里了"
python handler.py "我最近压力好大"
python handler.py "写租房合同用什么skill"
# Call as a module
from handler import handle
result = handle("我想查快递", installed_skills=["logistics"])
Data Files
data/skill-db.json- skill metadata database, including risk labelsdata/intent-map.json- intent keyword to recommended-skill mapping
Clawpilot v1.0.1 - the skill installation advisor, trust layer, and entry layer of ClawHub
安全使用建议
This skill appears coherent and low-risk: it runs locally (Python3) against the included JSON data files and only gives recommendations. Before installing, verify you are comfortable running the bundled Python script and optional test.js (the test uses node to spawn python3). Inspect the shipped data files (data/intent-map.json and data/skill-db.json) if you want to confirm the intent mappings and risk labels reflect your expectations. No credentials or network downloads are requested, and v1 explicitly does not auto-install other skills. If you need stronger guarantees, run the tests in an isolated environment (or inspect the code) and confirm your runtime has only trusted interpreters (python3/node).
功能分析
Type: OpenClaw Skill
Name: clawpilot-advisor
Version: 1.0.1
The 'Clawpilot' skill bundle is a recommendation engine designed to advise users on which other skills to install based on their intent. The implementation in 'handler.py' uses a simple keyword-matching logic against local JSON data files ('data/skill-db.json' and 'data/intent-map.json') and contains no network calls, file system modifications, or dangerous execution functions like eval() or os.system(). The documentation and code explicitly state and enforce that the skill provides advice only and does not perform automatic installations, and it includes proactive safety disclaimers for high-risk categories like legal and medical advice.
能力评估
Purpose & Capability
Name/description (installation advisor) align with provided files: a local intent map, a local skill DB, and a rule-based recommender. The skill does not request unrelated credentials or binaries and only needs a Python3 runtime as declared.
Instruction Scope
SKILL.md and handler.py limit behavior to: parse a user query, match keywords against the shipped intent map, consult the shipped skill DB, and generate a markdown report. Runtime instructions/CLI examples invoke handler.py locally; there are no instructions to read unrelated system files, call external endpoints, or exfiltrate data.
Install Mechanism
No install spec is provided (instruction-only). There are only local data files and scripts; nothing is downloaded or extracted from external URLs. This is the lowest-risk install model.
Credentials
The skill declares no required environment variables, no credentials, and no config paths. The code only opens the two local JSON data files. There is no disproportionate request for secrets or unrelated access.
Persistence & Privilege
The manifest does not request always:true and the skill does not modify other skills or system configuration. It is an on-demand advisor with no forced persistent privileges.
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 clawpilot-advisor - 安装完成后，直接呼叫该 Skill 的名称或使用
/clawpilot-advisor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
v1.0.1
Translate the published page documentation and summary to English while keeping the current advisor functionality and Chinese intent map unchanged.
v1.0.0
Initial release of Clawpilot - your Skill installation advisor.
- Analyzes user task intent and recommends suitable skills.
- Compares recommended skills, explains core differences, and risk levels (low, medium, high, pending).
- Offers installation order suggestions without automatic installation.
- Provides mandatory disclaimers for medium and high-risk skills.
- Handles unrecognizable or unsupported scenarios with guided responses.
元数据
常见问题
Clawpilot 是什么？
Clawpilot is your skill installation advisor. It analyzes task intent, recommends suitable skills, compares tradeoffs, explains risk, and suggests an install... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 112 次。
如何安装 Clawpilot？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install clawpilot-advisor」即可一键安装，无需额外配置。
Clawpilot 是免费的吗？
是的，Clawpilot 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Clawpilot 支持哪些平台？
Clawpilot 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 Clawpilot？
由 haidong（@harrylabsj）开发并维护，当前版本 v1.0.1。
推荐 Skills