A persistent context management system that uses a dedicated Markdown file to track active project state, preventing AI agents from losing track of their work.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install focus-tracker
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 focus-tracker 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 Focus Tracker?
The Focus Tracker is a vital component within the ecosystem of Openclaw Skills designed to solve the problem of context loss. AI agents often suffer from amnesia when a session restarts or when a context compaction event occurs, causing them to forget the current objective or specific sub-tasks. By maintaining a lightweight FOCUS.md file in the workspace root, this skill provides a source of truth that the agent can refer to immediately upon initialization.
It ensures that the agent always knows exactly what it was doing, why it was doing it, and what the very next step is without needing to ask the user for a status update. This persistent state management is a hallmark of high-quality Openclaw Skills that aim for professional-grade autonomy and reliability in complex development environments.
Focus Tracker Use Cases
- Resuming complex multi-step coding projects after an AI session restart or context compaction.
- Maintaining a clear roadmap for long-running tasks where the agent might otherwise drift toward irrelevant topics.
- Providing an immediate status update when a user asks what the agent is currently working on or where the project stands.
- Coordinating sub-agents by tracking their individual tasks and statuses in a centralized focus document.
- Transitioning between different high-priority tasks while ensuring previous work is properly archived and indexed.
How Focus Tracker Works
- At the start of every session or immediately after a context compaction, the agent checks for the existence of a FOCUS.md file.
- If the file exists, the agent reads the current objective, plan, and active state to resume work seamlessly without user intervention.
- As the agent completes sub-tasks or encounters blockers, it updates the FOCUS.md file in real-time to reflect the latest project status.
- Before any event that might flush the agent's short-term memory, it ensures the Active State section is updated with the most recent progress and next steps.
- Once a project is finished, the agent migrates the summary to a FOCUS-LOG.md file for long-term record-keeping and clears the active focus for the next task.
Focus Tracker Setup
To implement this within your collection of Openclaw Skills, ensure your agent is configured to check for the FOCUS.md file during its initialization routine. Add the following files to your workspace root:
touch FOCUS.md
touch FOCUS-LOG.md
Instruct your agent to read these files after reading its core identity and user preference files to ensure it has the most up-to-date context on current operations.
Focus Tracker Data Schema & Taxonomy
The system relies on two primary Markdown files to organize its state:
| File | Purpose | Key Metadata |
|---|---|---|
| FOCUS.md | Active task tracking | Project Name, Start Date, Status, Objective, Checklist, Active State, Blockers, Sub-Agents |
| FOCUS-LOG.md | History of completed work | Project Name, Completion Date, Duration, Final Outcome |
The FOCUS.md file is strictly kept under 50 lines to ensure it remains a high-density, efficient resume point for the agent context window.
Focus Tracker Advanced Features
- Multi-agent synchronization by listing spawned sub-agents and their specific statuses within the focus document.
- Automated archiving to FOCUS-LOG.md upon project completion to maintain a searchable history of accomplishments.
- Context compaction resilience by prioritizing the Active State section as the primary recovery point.
- Integration with session-start routines to enable proactive check-ins on unfinished work using Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
WeChat Remote Publisher
A remote-capable publishing tool that allows AI agents to post content to WeChat Official Accounts by bridging local requests through a fixed-IP MCP server.
Multi Search Engine
A powerful search integration for AI agents offering access to 17 global and domestic search engines without requiring API keys.
Profitabul MCP
A high-performance market intelligence skill for AI agents to analyze and execute futures and options trades with institutional-grade data.
Fast Claw Paid API
A specialized microservice skill for managing paid API access, credit top-ups, and persistent authentication for Fast Claw services.
agent-mineru
A high-performance document parsing CLI that transforms PDFs, Office files, and images into structured Markdown and JSON using advanced layout analysis.
CAD Skills
A specialized automation skill for controlling local Windows CAD applications including SolidWorks, CATIA, and Siemens NX through an AI agent.
Featured*
