A sophisticated tool designed to convert vague or underspecified user requests into clear, executable prompts with defined boundaries and acceptance criteria.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install prompt-refiner
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 prompt-refiner 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 Prompt Refiner?
The Prompt Refiner is a specialized utility for AI coding agents that bridges the gap between ambiguous human intent and precise technical execution. By analyzing session context and user goals, this tool restructures inputs to include explicit constraints, output formats, and verification steps. As an essential part of Openclaw Skills, it ensures that your AI agent understands the full scope of a task before writing a single line of code, significantly reducing the need for manual corrections.
Beyond mere wording polish, the skill focuses on task clarification and boundary setting. It works in a closed loop with local learning systems to adapt to specific user preferences over time. By incorporating this skill into your workflow, you ensure that every complex request is processed through a lens of technical rigor, leading to more predictable and high-quality outcomes.
Prompt Refiner Use Cases
- When user input is vague, lacks boundaries, or misses specific acceptance criteria.
- When a prompt needs to be reframed into a more executable format before work begins.
- Scenarios where a user explicitly asks to optimize, refine, or improve an existing prompt.
- Situations requiring substantial task clarification for multi-step engineering workflows.
How Prompt Refiner Works
- The skill extracts the original prompt and analyzes the current session context, including goals, tech stack, and expected outputs.
- It generates a refined version of the prompt, ensuring the core intent remains unchanged while making boundaries and requirements explicit.
- It determines the appropriate Continue Mode: Popup-confirm for manual review, Auto-apply for immediate execution, or Optimize-only for display.
- If refinement adds substantial value, it presents the optimized version to the user for confirmation.
- Upon user choice, it generates a learning signal to help the agent align with the user's habitual preferences for future interactions.
Prompt Refiner Setup
To integrate this skill into your agent environment, ensure it is registered within your toolset. Configuration typically involves setting the default continue mode.
# Example of how the skill is initialized within the environment
openclaw install prompt-refiner
The skill relies on access to the session context and works best when paired with a session-learner component to store preference patterns.
Prompt Refiner Data Schema & Taxonomy
The skill organizes refined data into several key blocks to ensure clarity and verifiability:
| Block Name | Purpose |
|---|---|
| Goal | Defines the primary objective of the request |
| Context | Includes tech stack, environment, and background |
| Constraints | Lists non-goals and specific limitations |
| Output Format | Specifies the exact structure of the desired result |
| Acceptance Criteria | Defines the verifiable points for task completion |
All preference data is stored as summarized rules (e.g., preference for seeing refined versions) rather than full prompt text to maintain privacy.
Prompt Refiner Advanced Features
- Smart value judgment: Automatically skips the refinement UI if the optimization doesn't add substantial functional value.
- Multiple Continue Modes: Flexible execution styles including Popup-confirm, Auto-apply, and Optimize-only.
- Preference Learning: Integrates with session-learner to capture user habits and refine its own optimization logic over time.
- Structured Heuristics: Uses specific patterns for code tasks versus content tasks to ensure domain-appropriate refinement.
- Openclaw Skills compatibility: Designed to work seamlessly with other agent protocols for holistic session management.
SKILL.md
Loading
Related Openclaw Skills
Anti-Crawler Evasion
A sophisticated toolkit designed to bypass anti-bot mechanisms through request spoofing, IP rotation, and human behavior simulation.
S2-Pet-Guardian
A professional Level 2 guardian agent providing digital twin monitoring, emotion translation, and secure IoT hardware orchestration for pet care.
OpenClaw Deployment: Alibaba Cloud ECS & DingTalk Integration
A comprehensive automation skill to deploy the OpenClaw AI agent platform on Alibaba Cloud ECS and integrate it with DingTalk bots.
CachyOS Expert
A senior systems engineer agent specialized in optimizing and troubleshooting CachyOS and Arch Linux environments.
Midas Skill
A self-learning wealth extraction engine that distills mundane daily information into actionable money signals and market opportunities.
Session Learner
An automated learning loop that captures stable user preferences and project conventions to continuously align AI behavior in future sessions.
Featured*
