# How to trace your vibe-coded agent with W&B Weave

## Executive summary

The video demonstrates how to implement comprehensive observability for AI agents using Weights & Biases (W&B) Weave and the W&B MCP server. By leveraging the `weave for agents SDK`, engineers can add full tracing—including conversations, turns, LLM calls, and tool executions—to an existing agent's logic without modifying its core code. This instrumentation allows developers to monitor performance metrics, track resource usage (tokens, cost), and debug complex interactions, such as identifying model hallucinations.

## Key takeaways

- Weave provides deep observability for AI agents: The tracing structure follows a clear hierarchy: Agent $\to$ Conversation $\to$ Turn $\to$ LLM Call + Tool Call. This detailed view is crucial for understanding agent behavior and performance.
- Non-invasive instrumentation using W&B MCP: Observability can be added by prompting a coding assistant (like Claude Code) to inject the necessary tracing logic via the `weave for agents SDK`, avoiding changes to existing application code.
- Debugging and Evaluation Capabilities: The Weave UI allows engineers to inspect individual conversations and turns, providing step-by-step visibility into tool usage (e.g., Tavali search) and LLM decisions. This is critical for debugging hallucinations or unexpected agent behavior.

## Technical details

- Weave Data Model Structure: The tracing model is hierarchical: Agent (top level) $\to$ Conversation (tied by ID) $\to$ Turn (user message + agent response) $\to$ LLM Call (single API hit) or Tool Call (single tool execution).
- Instrumentation Command Flow: The process involves prompting the coding assistant to add observability using the `weave for agents SDK` and specifying that each run should be a conversation, provider workflows as turns, and search/LLM calls as tool/LLM spans.
- W&B MCP Server Functionality: The W&B MCP server is used to integrate observability into the IDE environment, allowing the agent to pull traces and documentation directly within the development workflow.

## Practical implications

- Enables systematic evaluation of AI agents by providing granular data on every interaction (spans, tokens, cost).
- Facilitates debugging complex agent logic and identifying root causes for errors or hallucinations.
- Allows developers to compare performance metrics after making code changes by reviewing historical traces.

## Topics

AI Agents, Observability, LLM Tracing, Weights & Biases (W&B), Prompt Engineering, W&B MCP setup, Weave for Agents SDK

Source: https://www.youtube.com/watch?v=Mpa_oF6eh5w