A spec-driven development system that transforms complex project ideas into verified, working implementations through a structured Discuss-Plan-Execute-Verify workflow.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install gsd-claw
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 gsd-claw 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 GSD Claw?
GSD Claw is a high-performance implementation of the Get Shit Done (GSD) philosophy, specifically adapted for use with Openclaw Skills. Unlike standard chat-based interactions, this skill treats plans as executable prompts, ensuring that every coding task is grounded in clear specifications and measurable acceptance criteria. It is designed to handle substantial projects where structure and verification are more important than quick chat responses.
By prioritizing a solo developer + AI workflow, GSD Claw eliminates enterprise theater in favor of technical rigor. The system utilizes a fresh context pattern, spawning sub-agents for individual tasks to prevent context degradation over long development cycles. This makes it an essential tool for developers who want to maintain high code quality and architectural integrity while building complex features or entire applications.
GSD Claw Use Cases
- Planning and scoping new features or full-scale software projects from scratch.
- Executing complex technical tasks that require multi-step logic and automated verification.
- Managing project state and documentation across long-running development sessions.
- Transitioning from vague requirements to structured, executable XML-based task specifications.
- Automating bug fixes and dependency management through predefined deviation rules.
How GSD Claw Works
- Discuss Phase: The agent captures the user vision and identifies technical gray areas, documenting all decisions in a CONTEXT.md file to lock in project scope.
- Plan Phase: The system conducts technical research and creates plans consisting of 2-3 tasks, each defined using a strict XML structure with specific verification commands.
- Execute Phase: Tasks are processed by sub-agents to maintain fresh context windows, following deviation rules to automatically handle bugs or blockers while seeking permission for architectural changes.
- Verify Phase: Every deliverable is confirmed through automated command checks and a structured user acceptance walkthrough to ensure the implementation matches the original spec.
GSD Claw Setup
To begin using this system within your Openclaw Skills environment, navigate to your project directory and initialize the workflow with a natural language trigger. No complex manual installation is required beyond having the skill enabled in your agent configuration.
# Initialize a new project or feature scope
"Let's plan the [project name]"
# Or enter GSD mode directly
"GSD mode: build a secure authentication module"
The skill will automatically generate the .gsd/ directory structure and guide you through the initial Discuss phase.
GSD Claw Data Schema & Taxonomy
GSD Claw maintains a clean, transparent data taxonomy within the .gsd/ directory of your project to ensure all Openclaw Skills sessions stay aligned:
| File/Folder | Description |
|---|---|
PROJECT.md |
The high-level vision and core project goals. |
CONTEXT.md |
Locked decisions from the Discuss phase and areas of agent discretion. |
STATE.md |
Current position, phase, status, and tracked deviations. |
ROADMAP.md |
Outlines the phases of development and overall progress. |
plans/ |
Contains the individual task plans and execution result summaries. |
GSD Claw Advanced Features
- Fresh Context Pattern: Spawns sub-agents for execution to provide 200k context windows per task, preventing quality decay.
- Automated Deviation Handling: Systematically manages bugs and blocking issues without manual intervention, while pausing for architectural decisions.
- XML Task Schema: Uses a machine-readable XML format for tasks that includes specific paths, actions, and measurable 'done' criteria.
- Built-in Verification Loop: Rejects completion of tasks unless the defined verification commands or checks pass successfully.
- State Management: Persists project history and decision rationale across multiple sessions, ensuring continuity in long-term builds.
SKILL.md
Loading
Related Openclaw Skills
Frankenstein
A sophisticated synthesis engine that merges the strengths of multiple AI agents into a single, high-performance skill.
Bitrix24 AI Assistant
An intelligent interface for Bitrix24 that allows business owners to manage CRM, tasks, and team schedules using natural language.
Nanoleaf Lighting Control
Control Nanoleaf light panels via the Picoleaf CLI to manage power, brightness, and color settings programmatically.
PayRail402
A centralized financial control plane for tracking AI agent transactions across traditional and on-chain payment rails.
sandwrap
A protective wrapper that provides a soft-sandbox environment for untrusted skills using multi-layer prompt defense and tool restriction.
Token Saver
Token Saver is a model-aware optimization tool that reduces AI costs by dynamically compacting chat context and compressing workspace files based on specific model limits.
Featured*
