A sophisticated framework that transforms AI agents from reactive chatbots into proactive, context-aware copilots using persistent state management.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install copilot
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install copilot using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Copilot AI Skill?
The Copilot skill is designed to bridge the gap between intermittent agent activations and the need for continuous, real-time awareness. While AI agents typically operate in isolation across sessions, this skill enables a stateful memory system that allows the agent to fake continuity. By leveraging structured state files, the agent can remember ongoing tasks, previous decisions, and specific user preferences without requiring the user to provide repetitive context.
By integrating this into your collection of Openclaw Skills, you move away from generic chatbot interactions toward an opinionated, proactive partnership. The skill focuses on reducing the friction of fresh starts, ensuring that every time an agent wakes up—whether via a user message, a heartbeat poll, or a scheduled cron job—it already knows exactly where the work left off.
Copilot AI Skill Use Cases
- Resuming complex development tasks across different days without re-explaining context.
- Maintaining a persistent log of architectural decisions to ensure consistency in code generation.
- Proactive project switching where the agent automatically loads relevant documentation and history when changing directories.
- Background monitoring via heartbeats to offer timely assistance without being intrusive.
How Copilot AI Skill Works
- Initial Activation: On every trigger, the skill immediately reads the active context file to determine the current project, task, and potential blockers.
- Context Validation: The agent compares the stored state with the current environment. If the state is stale, it asks a targeted question; if fresh, it references the context naturally.
- Proactive Reasoning: Instead of offering generic options, the agent uses the stored patterns and priorities to provide opinionated recommendations.
- State Persistence: After any significant interaction or project switch, the skill updates the relevant Markdown files in the storage directory to ensure the next session is seamless.
- Optimization: The skill manages token costs by prioritizing file-based context over expensive visual screenshots.
Copilot AI Skill Setup
To implement this within your Openclaw Skills environment, create the required directory structure and initialize the base context files:
mkdir -p ~/copilot/projects
touch ~/copilot/active ~/copilot/priorities ~/copilot/decisions ~/copilot/patterns
Ensure your agent's system prompt is configured to read from ~/copilot/active as its primary source of truth during the initialization phase.
Copilot AI Skill Data Schema & Taxonomy
The skill organizes data within a dedicated directory structure for maximum portability and readability:
| File/Folder | Purpose | Update Frequency |
|---|---|---|
active |
Stores current focus, task, and immediate blockers. | Every activation |
priorities |
High-level projects, key stakeholders, and deadlines. | Weekly or on shift |
decisions |
An append-only log of technical or strategic choices. | Per significant decision |
patterns |
Learned user preferences, code styles, and shortcuts. | As patterns emerge |
projects/ |
Sub-directories containing project-specific state files. | On project switch |
Copilot AI Skill Advanced Features
- Heartbeat Intelligence: Differentiates between routine polling and urgent interruptions to minimize noise while maintaining awareness.
- Opinionated Guidance: Moves beyond the "How can I help?" anti-pattern to provide reasoned, proactive suggestions based on history.
- Cost-Aware Vision: Intelligently decides when to capture screenshots versus reading state files to optimize token usage.
- Multi-Project Lifecycle: Supports deep context switching that preserves the state of individual workspaces independently.
SKILL.md
Loading
Related Openclaw Skills
COO / Chief Operations Officer
Empower your AI agent to design repeatable processes, scale organizations, and drive cross-functional operational excellence.
Convex
A professional-grade skill for building, debugging, and scaling Convex backends with a focus on schema safety and operational reliability.
Contracts Skill
A secure, local system for tracking contract lifecycles, extracting key terms, and managing renewal deadlines automatically.
Contract
A structured AI skill for drafting contracts from scratch with guided discovery, risk analysis, and multi-party negotiation tracking.
CORS Configuration and Security
A comprehensive technical utility for correctly implementing, debugging, and securing Cross-Origin Resource Sharing (CORS) in web environments.
Cosmos
A comprehensive AI agent skill for navigating the Cosmos ecosystem, managing ATOM tokens, and executing cross-chain IBC transfers.
Featured*
