A strategic reasoning framework that forces AI agents to clarify assumptions and pressure-test plans before execution.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install socrates
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 socrates 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 Socrates?
Socrates is a sophisticated reasoning skill designed for AI coding agents within the Openclaw Skills ecosystem. It implements a structured Socratic self-Q&A protocol that ensures agents think before they act, significantly improving execution quality for non-trivial tasks. By exposing hidden assumptions and analyzing potential failure points, this skill transforms agents from reactive tools into proactive problem-solvers while avoiding the pitfalls of abstract philosophy or unnecessary latency.
Socrates Use Cases
- Architecture design and risky implementation decisions
- Complex refactoring with a broad blast radius
- Multi-agent delegation and workflow coordination
- Sequence design for long-running autonomous tasks
How Socrates Works
- The agent evaluates the task to determine if a full, compressed, or skipped pass is necessary based on rework risk.
- Execution of Round 1 focuses on identifying the real goal, surfacing unknowns, and evaluating alternative approaches.
- A conditional Round 2 analysis is triggered only for high-stakes architectural decisions or unresolved tradeoffs.
- The agent generates a structured Socratic Pass report to document its internal reasoning process.
- A final Action Plan is produced, including a Delegation Contract if sub-agents are required to complete the task.
Socrates Setup
To integrate Socrates into your workflow, ensure the skill definitions are accessible to your agent. This protocol is a core component of high-performance Openclaw Skills implementations.
# Example directory structure for Openclaw Skills integration
mkdir -p skills/socrates
# Ensure socrates/SKILL.md and references/ are properly mapped
Socrates Data Schema & Taxonomy
The Socrates skill organizes output into specific markdown sections to maintain consistency across Openclaw Skills integrations:
| Component | Type | Description |
|---|---|---|
| Socratic Pass | Section | Summary of the internal Q&A, risks, and approach choices |
| Action Plan | Section | The definitive, step-by-step technical roadmap |
| Delegation Contract | Section | Goal, done criteria, and constraints for sub-agents |
Socrates Advanced Features
- Intelligent workflow compression to ensure high-speed responses for trivial chat or minor edits
- Recursive delegation rules that require sub-agents to perform their own mini Socratic passes
- Built-in guardrails to prevent task stalling and ensure agents ask users for clarification when blocked
- Standardized handoff protocols for seamless multi-agent collaboration and clear accountability
SKILL.md
Loading
Related Openclaw Skills
Viral Article Deconstructor
A professional framework for deconstructing viral content into actionable writing strategies and emotional triggers.
Vinehoo Product Search & Statistics
A professional-grade wine discovery and statistical analysis tool for searching the Vinehoo database and tracking daily new arrivals.
AICodeM XMind Test Case Generator
Automatically convert structured test case data into professional XMind mind maps for enhanced visualization and quality assurance.
YouTube Super Chat Event
A streamlined skill for listing and managing YouTube Super Chat events through the yutu command-line interface.
YouTube Thumbnail
A specialized automation skill for managing and updating YouTube video thumbnails directly via the yutu CLI tool.
Multi-Agent Filesystem Governance
A governance framework for defining ownership, storage placement, and lifecycle rules for files managed by multiple AI agents.
Featured*
