Home > Tasks > Reasoning > ChainOfThoughtTask

ChainOfThoughtTask

Break down complex problems into explicit reasoning steps with per-step validation, confidence scoring, and automated backtracking.

Category: Reasoning Model: GPT-4 Preferred Side-Effect Safe
⚙️ TaskConfig.json

{
  "task_type": "ChainOfThought",
  "problem_statement": "Optimize cache invalidation in Auth.kt",
  "reasoning_depth": 5,
  "validate_steps": true,
  "related_files": ["src/**/Auth.kt"]
}
→
👁️ TabbedDisplay Output
Overview
Step 1
Step 2
Summary
### Step 1: Initial Analysis

Analyzing Auth.kt. The current TTL-based invalidation causes stale sessions during role updates.

Conclusion
Implement event-driven invalidation via onUserUpdate.
Confidence: 94.0%
Field Type Default Description
problem_statement * String - The complex problem requiring step-by-step reasoning.
reasoning_depth Int 10 Number of reasoning steps to generate (Range: 1-20).
validate_steps Boolean true Whether to validate each step before proceeding.
related_files List<String> [] Glob patterns (e.g. **/*.kt) for files to be used as input context.