A spec-driven development skill that enforces a structured, multi-phase workflow for AI coding agents to ensure high-quality feature implementation.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install spec-flow
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
Copy this prompt to OpenClaw to install it automatically.
Help me install spec-flow using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
Spec-Flow is a sophisticated development framework within the Openclaw Skills ecosystem designed to bring professional engineering discipline to AI-assisted coding. It moves away from chaotic code generation by enforcing a rigorous phase-by-phase progression: Proposal, Requirements, Design, Tasks, and finally Implementation. This skill ensures that your AI agent fully understands the scope and risks of a feature before writing any code, significantly reducing technical debt and logic errors.
By integrating Spec-Flow into your development cycle, you establish a clear trail of living documentation. The skill relies on mandatory user confirmation at every stage, ensuring the human developer remains in control of the architectural direction. Whether you are building complex systems or simple features, this addition to Openclaw Skills provides the structure necessary for reliable and scalable software development.
To begin using this skill in your Openclaw Skills environment, you can initialize a new feature workflow using the provided shell script or by direct prompt:
# Initialize a new feature workflow
./scripts/init-spec-flow.sh <feature-name>
You can also trigger it by telling your agent: "I need a plan for [feature]" or "Enter spec-flow mode for [feature description]". Ensure your project has a .spec-flow directory structure to store the generated documentation.
The skill organizes all development artifacts within a dedicated .spec-flow/ directory to maintain project cleanliness:
| Directory | Contents | Purpose |
|---|---|---|
steering/ |
constitution.md, tech.md |
Provides global project context and technical constraints. |
active/ |
proposal.md, requirements.md, design.md, tasks.md |
Contains the living documents for the feature currently in development. |
archive/ |
Completed feature folders | Stores historical records of finished implementations for future reference. |
Note: All generated markdown documentation is written in Chinese by default to support specific development team requirements.
--fast flag to generate all planning documentation in a single pass for smaller, well-defined tasks.--skip-design parameter for straightforward features where a full architectural document isn't necessary.Loading
A systematic framework for AI practitioners to transform information, research, and code changes into a structured, correlated technical knowledge map.

A high-level orchestrator that manages a team of specialized AI agents to handle the entire software development lifecycle from requirements to deployment.

A lightweight content archiving and review tool for IM workflows that saves links, notes, and images directly from chat.

A deterministic text analysis tool that decomposes documents into structured semantic units like risk, authority, and financial data without using an LLM.

A virtual crypto exchange simulation allowing AI agents to trade spot, perpetual futures with high leverage, and AMM swaps using a 10,000 USDT starting balance.

A professional management skill for monitoring and controlling Dell PowerEdge servers through the iDRAC Redfish API.








































