Chain of Thought

The Chain of Thought components at frontend/agentic-workbench/src/components/ChainOfThought/ provide LLM reasoning transparency by visualizing the step-by-step thought process of AI agents.

ChainOfThoughtViewer

File: frontend/agentic-workbench/src/components/ChainOfThought/ChainOfThoughtViewer.tsx

Renders the agent's reasoning chain as an expandable tree:

Each thought step shows the LLM's reasoning
Tool calls are displayed with inputs and outputs
Decision points highlight which branch was taken
Token counts per step
Latency per step
Confidence scores where available

Types

// frontend/agentic-workbench/src/components/ChainOfThought/types.ts
interface ThoughtStep {
  id: string;
  type: 'reasoning' | 'tool_call' | 'decision' | 'output';
  content: string;
  toolName?: string;
  toolInput?: Record<string, unknown>;
  toolOutput?: string;
  tokenCount: number;
  latencyMs: number;
  children?: ThoughtStep[];
}

Integration

The Chain of Thought viewer integrates with the AI Service's streaming response to render thought steps in real-time as the agent processes a request.

Rollback Controls Workflow Tracking