react-pattern

Thought-Action-Observation loop for transparent reasoning

170 stars

byMiosa-osa

View on GitHub Installation ↓

Best use case

react-pattern is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Thought-Action-Observation loop for transparent reasoning

Teams using react-pattern should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

You only need a quick one-off answer and do not need a reusable workflow.
You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/react-pattern/SKILL.md --create-dirs "https://raw.githubusercontent.com/Miosa-osa/canopy/main/library/skills/ai-patterns/react-pattern/SKILL.md"

Manual Installation

Download SKILL.md from GitHub
Place it in .claude/skills/react-pattern/SKILL.md inside your project
Restart your AI agent — it will auto-discover the skill

How react-pattern Compares

Feature / Agent	react-pattern	Standard Approach
Platform Support	Not specified	Limited / Varies
Context Awareness	High	Baseline
Installation Complexity	Unknown	N/A

Frequently Asked Questions

What does this skill do?

Thought-Action-Observation loop for transparent reasoning

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

# ReAct Pattern

Thought → Action → Observation loop for transparent reasoning and reduced hallucination.

## Activation

Activates on:
- Multi-step tasks
- Research tasks
- Debugging sessions
- Complex problem solving

## Process

### Loop Structure

```
THOUGHT: What do I need to do next?
ACTION: [tool_name] with [parameters]
OBSERVATION: [result of action]
... repeat until task complete ...
THOUGHT: Task complete because [reasoning]
```

### Guidelines

1. **Thought**: Explicit reasoning about next step
   - What information do I need?
   - What's the best tool for this?
   - What could go wrong?

2. **Action**: Execute one tool call
   - Use the most specific tool
   - Provide complete parameters
   - Prefer specialized tools over Bash

3. **Observation**: Analyze the result
   - Did I get what I expected?
   - Is there an error to handle?
   - What did I learn?

### Benefits

- **Transparency**: Reasoning is visible
- **Debuggability**: Each step is traceable
- **Reduced Hallucination**: Actions ground in reality
- **Learning**: Patterns can be extracted

## Integration

Works with:
- All agent workflows
- Memory system (captures thoughts)
- Learning engine (extracts patterns)
- Verification (traces reasoning)

## Anti-Patterns

Avoid:
- Skipping thoughts before actions
- Multiple actions without observations
- Ignoring error observations
- Concluding without final verification

---

*Based on ReAct: Synergizing Reasoning and Acting in Language Models*