← 返回 Skills 市场
210
总下载
0
收藏
0
当前安装
21
版本数
在 OpenClaw 中安装
/install setup-doctor
功能描述
Diagnose OpenClaw setup in one command. Auto-detects issues with Node, npm, gateway, config, and workspace. Quick mode (~10s) or Full mode. Multi-language. C...
安全使用建议
This skill appears to do what it says: local checks for Node/npm/OpenClaw and filesystem presence. Things to consider before installing: 1) The agent will run shell commands (node/npm/openclaw) to perform quick checks — ensure you trust running those binaries in your environment. 2) Auto-detection rules can cause the agent to run the quick check when you mention errors; if you prefer manual control, avoid enabling automatic invocation or confirm prompts. 3) The included script is local and does not perform network calls or read file contents, but there are small implementation bugs (e.g., minor path-handling issues) — expect possible noisy warnings rather than silent malicious behavior. 4) Always review and explicitly approve any “fix it” action the skill proposes before allowing it to execute changes.
能力评估
Purpose & Capability
The skill claims to diagnose OpenClaw/node/npm/gateway and its filesystems checks align with that purpose. It requests no credentials, no config paths, and includes a small local diagnostic script that inspects typical OpenClaw locations under the user's home directory — all proportional to a setup-diagnostic tool.
Instruction Scope
SKILL.md instructs the agent to run quick-check shell commands (node --version, npm --version, openclaw gateway status) via the agent's exec facility — this is expected for a diagnostic skill but means the agent will execute external binaries. The SKILL.md repeatedly states it will not read file contents; the included script only checks existence/access and does not read file contents. Minor mismatch: SKILL.md's Workspace Audit claims to report file sizes for specific files, but the included diagnose.js does not enumerate per-file sizes — an implementation gap rather than malicious behavior. Also: auto-detection triggers may cause the agent to run quick checks when users report problems; this is reasonable but the user should be aware of automatic invocation.
Install Mechanism
This is instruction-only (no install spec). A small local script is included but there is no download/install mechanism that would fetch remote code. Low install risk.
Credentials
The skill declares no required environment variables or secrets and the code accesses only local filesystem and process metadata (home directory, process.execPath, etc.). That access is proportionate to discovering host-side OpenClaw/Node/npm configuration. There are no network calls or child_process usage in the script.
Persistence & Privilege
always:false and no install actions that modify other skills or global agent config. The skill can be invoked autonomously per platform defaults; combined with its limited local-checks scope this is not excessive. Fix mode promises explicit confirmation before changes — users should confirm that in practice.
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 setup-doctor - 安装完成后，直接呼叫该 Skill 的名称或使用
/setup-doctor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
v3.0.1
Fixed config check wording (removed 'has required fields' that implied content reading).
v3.0.0
Auto-detect mode, simplified to Quick/Full. Removed over-engineered phases. Auto-triggers on setup issues.
v2.9.3
Install command now inline for visibility
v2.9.2
Added standalone install command
v2.9.1
Fixed contradictory privacy claim: 'Reads no personal files' replaced with accurate description. Config and workspace checks report file existence and size only. No content reading.
v2.9.0
Progressive disclosure: Quick Check (default, 10s) and Full Check (opt-in). Simplified from 4 sections to 2 modes. Removed Section D (generic safety rules). Cleaner description. Removed self-promotion section.
v2.8.0
Security: Fixed diagnose.js — removed package.json reading (now uses accessSync only). Fixed Node version check (>= 20, matches SKILL.md). Fixed misleading 'Invalid JSON' error message.
v2.7.0
Security: Phase 2 now checks config file existence only (Test-Path/test -f). Does NOT read config contents, API keys, or any values. Updated diagnose.js to use accessSync instead of readFileSync — no file content is read.
v2.6.0
Security: removed all USER.md references from file tables and templates.
v2.5.0
Security v5: Major rewrite. Removed ALL file reading (USER.md, memory files, workspace contents). Workspace audit now reports file existence/size only. Setup wizard provides templates instead of writing files. No credential storage. No USER.md access. Removed language preference persistence.
v2.4.0
Security: hardened secret handling language, added homepage/provenance.
v2.3.0
Security: replaced npx clawhub@latest with clawhub install in cross-promotion commands. No runtime behavior change.
v2.2.0
Added cross-promotion: recommends context-brief, locale-dates, tommo-skill-guard. Added starter pack install command.
v2.1.0
Optimized description with SEO keywords, cross-platform mention, expanded trigger phrases, cross-promotes tommo-skill-guard
v2.0.4
Minor metadata updates and workspace alignment
v2.0.3
Security hardening: Removed ALL child_process usage from diagnose.js. Script now uses only native Node.js APIs (fs, path, os, process) with zero shell execution. Added multi-path detection for OpenClaw binary across npm-global locations. Gateway check uses PID file detection; full gateway status delegated to agent exec tool.
v2.0.2
Security hardening: Rewrote diagnose.js to minimize child_process usage. Node.js and npm checks now use native process.versions and fs APIs (no shell execution). Only openclaw and gateway status still need CLI calls (no Node.js API equivalent). Added security header comment explaining remaining execSync usage.
v2.0.1
Security fix: API connectivity check is now opt-in only (requires 'doctor check api'). Default doctor check no longer touches API keys or sends authenticated requests. Network reachability test uses no credentials. Authenticated key test requires explicit user consent per-provider. Updated description to reflect security posture.
v2.0.0
Major update: Added workspace file audit with rules reference, agent permissions and security restrictions guide, initial setup wizard with 3 profiles (minimal/standard/power-user), file-by-file documentation, BOOTSTRAP.md cleanup, best-practice implementation with change reports. Cross-skill synergy with locale-dates, context-brief, and email-triage-pro.
v1.1.0
Added language preference system: 7 supported languages, persistent preference saved to USER.md, set language command, expanded output translations. Cross-platform fixes: PowerShell + bash commands.
元数据
常见问题
Setup Doctor 是什么？
Diagnose OpenClaw setup in one command. Auto-detects issues with Node, npm, gateway, config, and workspace. Quick mode (~10s) or Full mode. Multi-language. C... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 210 次。
如何安装 Setup Doctor？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install setup-doctor」即可一键安装，无需额外配置。
Setup Doctor 是免费的吗？
是的，Setup Doctor 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Setup Doctor 支持哪些平台？
Setup Doctor 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 Setup Doctor？
由 TommoT2（@tommot2）开发并维护，当前版本 v3.0.1。
推荐 Skills