← 返回 Skills 市场
spec-task
作者
LeoNardo-LB
· GitHub ↗
· v1.2.1
· MIT-0
172
总下载
0
收藏
0
当前安装
3
版本数
在 OpenClaw 中安装
/install spec-task
功能描述
结构化任务管理与生命周期强制执行。以下场景必须使用：(1) 任何被 coordinator 通过 sessions_spawn 派发的任务 (2) 可拆解步骤≥3 的复杂任务 (3) 工作区已存在 spec-task/ 目录时的后续任务 (4) 用户显式要求使用 spec-task (5) prependCont...
安全使用建议
This skill is coherent for enforcing a spec-driven task workflow and is instruction-only (no external downloads or credential requests). Before installing: (1) confirm you are comfortable with the skill creating and modifying files under your agent workspace (spec-task/<task>/, outputs/, memory/*), (2) ensure no sensitive secrets or PII will be written to those locations, (3) be aware child agents dispatched by a coordinator will be required to follow this workflow (may interfere with other agent behaviors), and (4) review and, if needed, adjust config.yaml (timeouts, auto_archive, per_agent_archive) to match your environment. A minor oddity: docs mention legacy .py script names even though no Python scripts are included—this appears to be a documentation remnant only.
功能分析
Type: OpenClaw Skill
Name: spec-task
Version: 1.2.1
The spec-task skill bundle is a structured task management framework designed to enforce a rigorous lifecycle (briefing, planning, execution, and verification) for AI agents. It manages task states via local YAML and Markdown files (e.g., status.yaml, checklist.md) and provides tools for logging and archiving project history. There is no evidence of data exfiltration, unauthorized command execution, or malicious intent; the prescriptive instructions in SKILL.md and schema.yaml are solely focused on ensuring agent compliance with the defined organizational workflow.
能力评估
Purpose & Capability
Name and description (structured task management, lifecycle enforcement) match the provided artifacts: SKILL.md, schemas, templates, and config all define a spec-driven task workflow. There are no unrelated required env vars, binaries, or install steps.
Instruction Scope
The runtime instructions require reading and writing workspace files (spec-task/<task>/status.yaml, brief/spec/plan/checklist, outputs/, memory/task-history, etc.) and mandate a strict workflow for child agents. That file I/O and workspace mutation are expected for a task-management skill, but you should be aware it will create and update files in the agent workspace and agent memory. There are no instructions to call external network endpoints or access unrelated system config.
Install Mechanism
Instruction-only skill with no install spec and no code files to fetch or execute. This is the lowest-risk install profile.
Credentials
The skill declares no required environment variables, credentials, or config paths. The config.yaml controls local behavior (timeouts, archive rules). Requested accesses (workspace, writing memory archives) are proportional to a task-management capability.
Persistence & Privilege
always:false (normal). The skill expects to write archival history and lessons into agent memory and workspace (memory/task-history, memory/task-lessons, outputs/). This is typical for a task-tracking skill but means artifacts and potentially sensitive content could be stored in the agent workspace/memory—review storage policies if that is a concern. The skill also enforces use by spawned child agents per its instructions; that is a behavioral constraint rather than an elevated system privilege.
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 spec-task - 安装完成后，直接呼叫该 Skill 的名称或使用
/spec-task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
v1.2.1
The skill has been deprecated due to unstable factors and can be used in the form of a plugin with the same name instead.
v1.2.0
**Summary:**
Introduced a historical recall feature and made configuration/usage scope clearer.
- Added `task_recall.py` script for retrieving relevant task history and lessons before starting new tasks.
- Updated documentation to require running recall before task creation to promote reuse of prior work and avoid repeated mistakes.
- Adjusted config and file paths: now use `spec-task/` as the standard task and config directory instead of `spec-tasks/`; updated related documentation and examples.
- Added a metadata section specifying Python3 as a runtime requirement.
- Slightly reduced test coverage notice from 117 to 115 case references.
- Clarified that the skill is for tasks with 5 or more steps and is intended for complex/structurable tasks.
v1.0.0
Spec-task 1.0.0 – Initial Release
- Introduces a spec-driven agent task management system using a structured workflow: brief → spec → plan → checklist → verify → archive.
- All task state managed by scripts and stored in files (status.yaml); agents focus on generating creative content.
- Fully automates config initialization and deep merge from project and skill defaults.
- Provides robust scripts for task creation, state transitions, checklist management, verification, archiving, and recovery.
- Enables traceable revisions, resumable checkpoints, configurable failure policies, and persistent project/task histories.
- All core operations documented, with user-friendly command references and flexible error handling workflows.
元数据
常见问题
spec-task 是什么？
结构化任务管理与生命周期强制执行。以下场景必须使用：(1) 任何被 coordinator 通过 sessions_spawn 派发的任务 (2) 可拆解步骤≥3 的复杂任务 (3) 工作区已存在 spec-task/ 目录时的后续任务 (4) 用户显式要求使用 spec-task (5) prependCont... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 172 次。
如何安装 spec-task？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install spec-task」即可一键安装，无需额外配置。
spec-task 是免费的吗？
是的，spec-task 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
spec-task 支持哪些平台？
spec-task 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 spec-task？
由 LeoNardo-LB（@leonardo-lb）开发并维护，当前版本 v1.2.1。
推荐 Skills