← 返回 Skills 市场
Project Context Manager
作者
liuboyang025-sketch
· GitHub ↗
· v1.0.0
· MIT-0
158
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install openclaw-project-context-manager
功能描述
用统一规则管理长期项目的发现、进入、退出、恢复与阶段性固化。
安全使用建议
This skill appears to do exactly what it says: manage project context via files and initialize project folders. Things to consider before installing or running:
- Review the included templates and scripts locally (scripts/init_project.py) so you understand exactly what will be written and where. The init script uses the projects_root you pass and defaults to creating a workspace under ~/.openclaw/workspace.
- The skill will read and write files under user-specified paths; only run initialization after confirming the target paths and backing up any existing data you care about.
- The docs contain example absolute paths (e.g. /Users/example/...). These are benign examples but you may want to remove or replace them before publishing or sharing the skill to avoid leaking local paths.
- There are no credential or network requests in the bundle; if you modify the skill to add remote sync or integrations later, re-evaluate required permissions.
If you want extra caution: run the provided init_project.py manually from a sandbox directory first (pass a temporary projects_root) to see the exact files created, and confirm the assistant asks for explicit confirmation before creating or registering projects as documented.
功能分析
Type: OpenClaw Skill
Name: openclaw-project-context-manager
Version: 1.0.0
The skill is a comprehensive project context management system designed to organize long-term AI interactions into structured file-based workspaces. It utilizes a Python script (scripts/init_project.py) to automate the creation of project directories and standardized markdown templates (e.g., recovery entries, decision logs, and checkpoints). The instructions in SKILL.md are well-structured and emphasize safety, requiring the agent to seek user confirmation before switching contexts and maintaining clear boundaries between project-specific data and general conversation. No malicious patterns such as data exfiltration, unauthorized network access, or obfuscated code were detected.
能力评估
Purpose & Capability
Name/description describe file-based project discovery, enter/exit/recover/solidify flows. Declared dependency is python3 and the bundle includes a small init_project.py plus templates and docs — all proportional to initializing and managing on-disk project structures.
Instruction Scope
SKILL.md instructs the assistant to read/write a lightweight projects registry (recommended path: ~/.openclaw/workspace/projects-registry.json), scan user-specified project directories, and read project files (00_恢复入口.md, indexes, checkpoints). That is consistent with the stated purpose. Note: the skill will create files/directories when the init script is run; SKILL.md and docs emphasize not writing paths into the skill body and require user confirmation before creating directories, but the included docs contain example absolute user paths (e.g. /Users/example/...) — harmless for functionality but should be removed before public release and may confuse reviewers.
Install Mechanism
No remote install or downloads; this is instruction-only with an included Python script. No network fetches or packaged dependencies were specified. Low install risk.
Credentials
The skill requires no environment variables or credentials. It does read/write user filesystem locations (user-specified projects_root and defaults under ~/.openclaw/workspace). This file-system access is necessary for the declared functionality and is limited to local paths the user supplies or the documented defaults.
Persistence & Privilege
The skill is not always-enabled and does not request elevated privileges. It can create and write files within the user's chosen project_root and workspace_root (defaulting to ~/.openclaw/workspace) when the init script is executed — expected for this behavior. Autonomous model invocation is allowed by default but is not exceptional here; the skill does not 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 openclaw-project-context-manager - 安装完成后,直接呼叫该 Skill 的名称或使用
/openclaw-project-context-manager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
v1.0.0
Initial release: establishes a universal, file-based long-term project context management system for multi-project workflows.
- Defines workspace-level rule files (e.g., PROJECT_SYSTEM.md) and per-project standard structures, supporting explicit project enter, recover, and checkpoint actions.
- Formalizes processes for project discovery, confirmation, entry, exit, and state recovery; prevents accidental mixing of project context and chat.
- Introduces standard project folder templates, including recovery entry, documentation index, checkpoints, and decision logs.
- Provides initialization script for creating compliant project structures.
- Sets out clear boundaries: project context is managed by files, distinct from UI session switching, and without hardcoded project assumptions.
- Establishes strict anti-interruption and prompt rules, ensuring user intent always controls project management workflows.
元数据
常见问题
Project Context Manager 是什么?
用统一规则管理长期项目的发现、进入、退出、恢复与阶段性固化。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 158 次。
如何安装 Project Context Manager?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install openclaw-project-context-manager」即可一键安装,无需额外配置。
Project Context Manager 是免费的吗?
是的,Project Context Manager 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Project Context Manager 支持哪些平台?
Project Context Manager 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 Project Context Manager?
由 liuboyang025-sketch(@liuboyang025-sketch)开发并维护,当前版本 v1.0.0。
推荐 Skills