The Complete Production SOP: How to Deploy and Run Weights & Biases Weave LLM Prompt Logger in Autonomous AI Agents

---
name: "weights-biases-prompts-logger"
description: "Log, version, and evaluate LLM inputs, outputs, and dataset iterations with W&B Weave."
metadata:
  author: "Weights & Biases"
  version: "1.0.0"
  runtime_compatibility:
    - "codex"
    - "claude-code"
    - "cursor"
    - "windsurf"
triggers:
  - "log prompts with wandb weave"
  - "version prompt templates weave"
  - "track llm experiments weave"
permissions:
  fs_read: true
  fs_write: "workspace"
  shell_exec: false
  net_access: false
---

import os
import weave
from dotenv import load_dotenv

load_dotenv(".env.local")

# Initialize trace project with Weave
weave.init(project_name=os.getenv("WEAVE_PROJECT", "agent-prompt-evals"))

@weave.op()
def generate_rag_response(user_query: str, system_context: str) -> dict:
    # Simulated LLM generation block with traced inputs and outputs
    payload = {"role": "system", "content": system_context}
    response = f"Synthesized response for: {user_query}"
    return {"query": user_query, "system_context": system_context, "output": response}

import pytest
from llm_service import generate_rag_response

def test_weave_tracing_contract():
    res = generate_rag_response("test prompt", "system prompt")
    assert "output" in res
    assert res["query"] == "test prompt"