Command Center
Overview of traces, spend, and live service health across your projects.
Observability for every LLM call your app makes
Trace prompts, completions, latency, tokens, and cost across OpenAI, Anthropic, and OpenRouter with a single SDK for TypeScript and Python.
Quick start
Install Pulse server first
Run this script to install the Pulse server binary, then start instrumenting with the SDK.
Server install script
curl -fsSL https://raw.githubusercontent.com/EK-LABS-LLC/trace-service/main/scripts/install.sh | bash -s -- pulseIntegrations
Works with the agents your team already uses
Native integrations for agent runtimes and CLIs so traces, runs, and tool activity land in one system.
Dashboard
See exactly what your team will use day to day
Fast visibility for tracing, run replay, triage, and spend decisions.
Agent Run Playback
Replay each agent run with tool calls, timing, and failure points in order.
Cost Intelligence
Track token usage and cost trends to catch spikes before they become problems.
Features
Everything you need to understand your LLM stack
From individual traces to full session timelines, Pulse gives your team the visibility to ship with confidence.
Tracing
End-to-end trace capture
Every prompt, completion, token count, latency, and cost stored with full request and response bodies.
Sessions
Session timelines
Group related calls into sessions. See full conversations or agent runs as a timeline.
Analytics
Cost and usage tracking
Track spend per model, per project, per session. Catch cost spikes before they hit your bill.
Routing
Provider agnostic
OpenAI, Anthropic, OpenRouter. One SDK, consistent metrics across all providers.
Debugging
Fast root-cause workflows
Pivot from aggregate analytics to individual traces and spans quickly when incidents happen.
Deployment
Single-node to scale mode
Run as a single binary with built-in SQLite, or switch to Postgres + partitioned listeners for higher throughput.
How it works
Instrument in minutes, not days
Three steps to full observability across your LLM stack.
Step 01
Install the SDK
Add the SDK to your project. Works with JS/TS and Python runtimes.
bun add @pulse/sdk
pip install pulse-sdkStep 02
Wrap your LLM client
Call observe() on your existing client. No API rewrite.
const client = observe(
new OpenAI({ apiKey: "sk-..." }),
Provider.OpenAI
);Step 03
Open the dashboard
Traces flow in automatically. Search, filter, and analyze quickly.
142 traces captured
$0.47 total cost
avg latency 1.2sDocumentation
Everything you need to track your agent usage
Guides, API reference, and examples to get from zero to production.
Getting started
Quickstart
Install the SDK, create an API key, and capture your first trace in under 5 minutes.
Read guide →SDK
Provider adapters
Integrate with OpenAI, Anthropic, and OpenRouter using the observe() wrapper.
Read guide →API
REST API reference
Query traces, sessions, and analytics programmatically. Full endpoint documentation.
Read guide →Concepts
Sessions & metadata
Group traces with session IDs and attach custom metadata for filtering.
Read guide →Config
Configuration
Batching, flush intervals, disabling tracing, and all initPulse() options.
Read guide →Start tracking your agents
Free and open source. Self-host anywhere.