← 返回 Skills 市场
221
总下载
1
收藏
1
当前安装
3
版本数
在 OpenClaw 中安装
/install tasker
功能描述
Use for task execution, debugging, implementation, analysis, review, planning, workflow execution, and user dissatisfaction handling in agent interactions. 任...
安全使用建议
This skill is coherent: it describes a workflow controller that requires actual tool calls to perform side effects and verification. Because it mandates executing platform tools (shell, read/write, etc.) when those are available, only install it if you trust the agent runtime and its tool permissions. Confirm what concrete executor APIs/permissions (shell access, file write/read, network) the platform will allow the skill to use, test the skill in a low-privilege environment first, and verify logging/auditing so you can see any actions it takes. If provenance of the skill (owner, homepage, source) matters to you, ask the publisher for more information before enabling it in a high-privilege context.
功能分析
Type: OpenClaw Skill
Name: tasker
Version: 1.1.1
The 'tasker' skill is a meta-prompting framework designed to enhance the reliability and transparency of AI agent workflows. It enforces strict operational constraints, such as 'Zero-Tool Zero-Progress' and 'Post-Tool Verification,' which require the agent to provide verifiable artifacts (e.g., command output or file reads) before claiming task completion. The skill includes a risk-based sizing mechanism (S/M/L) that mandates higher levels of validation for sensitive operations like database or production environment changes, and it provides structured protocols for handling user dissatisfaction. No malicious logic, data exfiltration, or unauthorized execution patterns were found in SKILL.md or _meta.json.
能力评估
Purpose & Capability
The name/description (general task execution, debugging, planning, review) match the SKILL.md instructions. The skill requests no environment variables, binaries, or config paths and has no install steps, which is proportionate for an instruction-only workflow controller.
Instruction Scope
The SKILL.md explicitly requires and enforces the use of executive tool calls (WriteFile, Shell, ReadFile, etc.) for any external side effects and mandates artifacts and verification. That behavior is coherent for a task executor but means the agent will attempt real side effects when platform tools are available — the file does not instruct reading unrelated system secrets or arbitrary external endpoints.
Install Mechanism
No install spec and no code files — instruction-only. There is nothing downloaded or written to disk by the skill itself.
Credentials
The skill declares no required env vars, credentials, or config paths. The runtime instructions do not request extra secrets or unrelated credentials.
Persistence & Privilege
always is false, user-invocable is true, and model invocation is enabled (platform default). The skill does not request permanent presence or claim to modify other skills or system-wide settings.
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 tasker - 安装完成后，直接呼叫该 Skill 的名称或使用
/tasker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
v1.1.1
Version 1.1.1 — Introduces strict execution safeguards to enforce real tool usage and verifiable changes for task steps.
- Added "Execution Guarantees" section: mandates that all executive steps require real tool calls, verifiable artifacts, and post-action verification.
- Agents are now prohibited from outputting completion statements before tool actions are confirmed and results returned.
- Every executive sub-step must produce a checkable artifact; otherwise, it remains in a `[pending]` state.
- Tool-Call Audit is now required in the verification phase to explicitly map all execution tools to deliverables.
- Clarifies that if an external change can't be made due to missing tools, execution is blocked and clarified to the user.
v1.1.0
**Tasker v1.1.0 - Risk-Intent Workflow**
What's new:
- **Signal weights**
: Time-based → Intent-based sizing ("view production logs" = high risk)
- **Dynamic clarify**
: Only ask for missing fields
- **Understanding check**
: M/L-level reverse confirmation before execute
- **External validation**
: L-level requires human/tool verification, no self-check
- **Auto-close**
: No response = accepted, non-blocking
Breaking: Fast-path removed. All levels require confirmation.
v1.0.0
Initial release of Tasker. Adds a general task execution skill with lightweight task handshake, AI-first sizing, concise output rules, review mode, and optional PUA integration.
元数据
常见问题
Tasker 是什么？
Use for task execution, debugging, implementation, analysis, review, planning, workflow execution, and user dissatisfaction handling in agent interactions. 任... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 221 次。
如何安装 Tasker？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install tasker」即可一键安装，无需额外配置。
Tasker 是免费的吗？
是的，Tasker 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Tasker 支持哪些平台？
Tasker 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 Tasker？
由 tealun（@tealun）开发并维护，当前版本 v1.1.1。
推荐 Skills