An advanced autonomous mode for AI agents that prioritizes codebase reading and self-verification over user questioning.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install just-keep-working
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 just-keep-working 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 OpenClaw Autonomous Programming?
OpenClaw Autonomous Programming is a specialized logic layer designed to transform how AI agents interact with complex codebases. Instead of relying on constant user clarification, this skill empowers agents to resolve uncertainties by actively reading existing code, identifying patterns, and maintaining a strict discipline of focus. It is a vital component for anyone looking to extend their library of Openclaw Skills with high-reliability automation.
At its core, the skill enforces a Sobriety Protocol, ensuring the agent never builds upon unverified or hallucinated progress. By distinguishing between additive and destructive changes, it maintains project integrity while executing multi-step tasks. This skill ensures that the agent adheres to the Project Constitution—the set of established naming conventions, architectural patterns, and styling rules already present in your repository.
OpenClaw Autonomous Programming Use Cases
- Executing large-scale feature additions across multiple files without manual guidance.
- Performing complex refactors while strictly adhering to existing project conventions.
- Automating bug fixes that require deep dependency tracing and self-verification.
- Implementing new UI components that must match the established styling and state management patterns.
How OpenClaw Autonomous Programming Works
- The agent scans the codebase to answer its own technical questions regarding frameworks and dependencies.
- A comprehensive Task Tree is generated, categorizing every planned action as either Additive or Destructive.
- The agent executes tasks foundational to the request, such as updating types, utilities, or configurations.
- After every file modification, a Sobriety Check is performed by re-reading the file to verify the output.
- A final Code Read-Back is conducted to ensure no placeholder comments or broken imports remain.
- A detailed completion report is provided, marking every task as verified and integrated.
OpenClaw Autonomous Programming Setup
To integrate this autonomous capability into your environment, add the skill definition to your agent's configuration. Use the following command to initialize the directory if you are managing Openclaw Skills via CLI:
mkdir -p .openclaw/skills
cp openclaw-autonomous.md .openclaw/skills/
Ensure your agent profile is configured to trigger this skill for all code-related requests.
OpenClaw Autonomous Programming Data Schema & Taxonomy
The skill organizes its execution logic through a structured task management system and internal logs:
| Data Component | Description |
|---|---|
| Task Tree | A hierarchical list of READ, ADD, EDIT, and DELETE actions. |
| Sobriety Log | Tracks task states: [DONE — VERIFIED], [DONE — UNVERIFIED], or [PENDING]. |
| Project Constitution | A derived set of rules based on the existing file structure and naming conventions. |
| Read-Back Checklist | A validation manifest used to review code before final submission. |
OpenClaw Autonomous Programming Advanced Features
- Non-Destructive Default: Automatically pauses and seeks confirmation before any file deletions or schema changes.
- Full-Scope Task Tree: Generates a complete dependency map before touching a single file.
- Hallucination Prevention: The Sobriety Protocol forces the agent to re-read files it just wrote to confirm they exist as expected.
- Contextual Architecture Scanning: Automatically identifies and matches project-specific state management and styling systems.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Autonomous Programming
A high-autonomy protocol that transforms AI agents into senior engineers capable of executing complex coding tasks from plan to verification without interruption.
klawdin
An AI-driven networking skill that allows agents to represent their owners on KlawdIn through automated profile management and active social engagement.
X Search
A powerful AI agent skill for real-time X/Twitter search, sentiment analysis, and social media intelligence powered by the x402 payment protocol.
Find People OSINT
An AI-powered OSINT tool designed for professional background research, due diligence, and career timeline verification using the x402 protocol.
InstantDB Integration
A powerful Node.js skill for Openclaw that enables real-time database administration and live data synchronization using InstantDB.
APort Agent Guardrail
A deterministic pre-action authorization engine that verifies AI agent tool calls against a security policy before execution.
Featured*
