← 返回 Skills 市场
270
总下载
0
收藏
2
当前安装
11
版本数
在 OpenClaw 中安装
/install project-cog
功能描述
AI project management powered by CellCog. Knowledge workspaces, document upload, AI-processed context trees, signed URL retrieval. Works standalone or as Cel...
安全使用建议
This skill appears to be a legitimate CellCog client wrapper, but it allows agents to upload any local files to CellCog using your CELLCOG_API_KEY. Before installing: (1) Confirm you trust CellCog (https://cellcog.ai) and understand their data handling policy. (2) Treat the CELLCOG_API_KEY like a secret — consider scoping or rotating the key and avoid using a high-privilege key. (3) If the platform lets you disable autonomous invocation for this skill, consider doing so or require manual confirmation before uploads. (4) Do not give the skill access to sensitive directories, or run it in a sandboxed environment when first testing. (5) Ask the publisher to mark CELLCOG_API_KEY as the primary credential in metadata to ensure correct handling by the platform. If you need help assessing data-sensitivity or configuring least-privilege keys, take those steps before enabling the skill.
功能分析
Type: OpenClaw Skill
Name: project-cog
Version: 1.0.10
The project-cog skill is a legitimate integration for the CellCog AI platform, providing tools for document management, project organization, and context-aware chat. The SKILL.md documentation correctly describes the use of the 'cellcog' Python dependency and the cellcog.ai API for managing knowledge workspaces, with no evidence of malicious intent, data exfiltration beyond the stated service, or prompt injection attacks.
能力标签
能力评估
Purpose & Capability
The name/description, required binary (python3), and required env var (CELLCOG_API_KEY) align with the SKILL.md which shows using the CellCog Python SDK to create projects, upload documents, and retrieve context trees. The requested artifacts are sensible for a CellCog integration.
Instruction Scope
Runtime instructions explicitly tell agents to upload local files (file_path arguments like /data/...), poll local document states, and retrieve signed URLs. Uploading arbitrary local files to an external API is within the claimed purpose (document management) but also enables exfiltration of sensitive local data if misused. The guidance does not impose path or file-type restrictions beyond a broad supported-file list; there is no caution about sensitive data.
Install Mechanism
This is an instruction-only skill with no install spec and no downloaded code. The metadata references a Python dependency 'cellcog' but there is no URL download or archive extraction. That is low-risk from an installation/execution perspective, though it assumes the runtime environment already provides python3 and the cellcog package.
Credentials
Requesting CELLCOG_API_KEY is proportionate to a service that requires authenticated API access. However, the registry metadata lists no primary credential despite requiring this API key, which is an inconsistency that could affect how the platform protects or displays the secret. No other unrelated credentials are requested.
Persistence & Privilege
The skill does not set always:true (good) and uses standard autonomous invocation (disable-model-invocation=false). Autonomous invocation combined with instructions that permit uploading arbitrary local files and using the user's API key increases the risk of unintended data exfiltration if the agent runs tasks without careful user prompts or constraints.
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 project-cog - 安装完成后，直接呼叫该 Skill 的名称或使用
/project-cog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
v1.0.10
- Updated the skill description to clarify its focus on AI project management and polished feature details.
- Revised the usage section to clarify agent-specific instructions, especially distinguishing OpenClaw from other agents.
- Minor improvements and clarifications throughout SKILL.md for better guidance and readability.
v1.0.9
- Updated documentation in SKILL.md for improved clarity and consistency.
- Added explicit agent provider initialization examples to usage instructions.
- Refined skill description to clarify features and available workflows.
- No code or interface changes; documentation only.
v1.0.8
project-cog v1.0.8
- Updated SKILL.md with improved usage instructions and guidance.
- Added a "How to Use" section clarifying agent integration examples.
- Emphasized referencing the main cellcog skill for detailed SDK documentation.
- No code or API changes.
v1.0.7
- Changed Quick Start instructions to use agent_provider="openclaw" instead of agent_name="openclaw" for CellCogClient initialization.
- No other functional or documentation changes.
v1.0.6
- Added agent_name="openclaw" to the CellCogClient example in the "Quick Start" section for clearer agent specification.
- No functional or API changes; documentation (SKILL.md) only.
- Improves clarity for users integrating with OpenClaw agent workflows.
v1.0.5
project-cog 1.0.5
- Major SKILL.md overhaul: rewritten and restructured documentation for clarity and usability.
- Added clear explanations of CellCog Projects, usage modes (with chats or standalone), and project/document lifecycles.
- Provided detailed, code-focused quickstart and practical workflow examples.
- Clarified the distinction between project_id and context_tree_id in API calls.
- Expanded information on document processing, context tree usage, and file type support.
- Cleaned up and condensed the feature/usage list for streamlined onboarding.
v1.0.4
project-cog v1.0.4
- Refined SKILL.md for clarity and conciseness.
- Improved explanation of usage modes (CellCog chats vs. standalone agent memory).
- Simplified setup and prerequisites; points users to cellcog mothership skill for detailed SDK instructions.
- Highlighted key features: persistent context, document upload, AI-generated context trees, and multi-agent access.
- Added overview of related skills for ecosystem context.
v1.0.3
- Updated the Quick Start guide in SKILL.md to clarify usage of create_chat() for OpenClaw agents (using notify_session_key) versus other agents.
- No code or functionality changes; only documentation updated.
- Helps users distinguish agent-specific integration patterns when creating chats with project documents.
v1.0.2
- Expanded standalone usage: clarified that any agent (not just OpenClaw) can use CellCog's Context Trees.
- Updated Chat usage example: simplified `create_chat()` parameters and noted it now blocks until done, returning full results.
- Minor documentation clean‑ups and improved language for clarity.
- No changes to code or functionality; documentation only.
v1.0.1
- Metadata in SKILL.md updated for clarity and standardization.
- Added a homepage field with the CellCog website.
- Improved structuring of requirements and environment fields under metadata.
- No changes to Python API or usage instructions.
v1.0.0
CellCog Projects 1.0.0 — initial release for knowledge workspaces.
- Create and manage AI-powered knowledge workspaces with project-based access.
- Upload documents of various types; receive structured, hierarchical context tree summaries processed by CellCog's AI.
- Retrieve signed URLs for document sharing.
- Works standalone for any agent, or as context for CellCog chat agents.
- Granular admin controls for managing projects and documents.
- Supports polling for document processing and exports context trees in markdown for easy agent consumption.
元数据
常见问题
Project Cog 是什么？
AI project management powered by CellCog. Knowledge workspaces, document upload, AI-processed context trees, signed URL retrieval. Works standalone or as Cel... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 270 次。
如何安装 Project Cog？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install project-cog」即可一键安装，无需额外配置。
Project Cog 是免费的吗？
是的，Project Cog 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Project Cog 支持哪些平台？
Project Cog 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（darwin, linux, windows）。
谁开发了 Project Cog？
由 CellCog（@nitishgargiitd）开发并维护，当前版本 v1.0.10。
推荐 Skills