Cookbook
Copy-paste recipes for common 2Signal tasks. Each recipe is a self-contained guide you can follow start to finish — pick the one that matches what you are trying to do and have it running in minutes.
Hello World Tracing
Get traces flowing in under 5 minutes with Python, TypeScript, or Go.
Auto-Instrument LLM Calls
Wrap OpenAI, Anthropic, and other providers for automatic tracing.
Evaluate Agent Outputs
Set up evaluators to automatically score every trace.
Validate Output Format
Use JSON Schema and Regex evaluators to enforce structured outputs.
Monitor Cost & Latency
Track spend and response times with automatic alerts.
Route Models by Complexity
Send simple queries to cheap models and complex ones to powerful models.
Version-Control Prompts
Manage prompt templates with immutable versions.
Run Evaluations in CI/CD
Block deploys when agent quality regresses.
Trace Replay & A/B Testing
Re-run traces with different models or prompts and compare results.
OpenTelemetry Integration
Send OTLP traces to 2Signal from any OTel-instrumented app.
Safety Evaluators
Set up bias detection, PII detection, toxicity, and prompt injection monitoring.
Webhook Evaluator
Build a custom evaluation endpoint and connect it to 2Signal.
A/B Test Prompts
Compare prompt variants with traffic splitting and statistical significance.
Human Review & Labeling
Set up review queues, label traces, and export to datasets.