latencyx 0.2.0

Zero-config latency tracking for FastAPI (and later more)

LatencyX

Lightweight auto-instrumentation for Python web apps. Because OpenTelemetry made you question your life choices.

Built for solo developers and small teams who want observability without needing a PhD in distributed systems.

Why LatencyX?

• One-line setup - latencyx.init(app) and you're done
• Auto-instrumentation - FastAPI, HTTP clients (httpx). More coming.
• Multiple exporters - Console, JSON files
• Minimal dependencies - Won't break your deployment
• Actually simple - No 50-page configuration docs

Quick Start

pip install latencyx

from fastapi import FastAPI
import latencyx

app = FastAPI()
latencyx.init(app)  # That's it.

@app.get("/")
async def root():
    return {"hello": "world"}

Output:

INFO:latencyx:[http.server] GET / duration=50.58ms status=200 method=GET client=127.0.0.1 path=/

CLI Monitoring

Watch your traces in real-time with a pretty CLI:

latencyx tail

📊 Watching LatencyX traces from: latencyx_traces.jsonl
   Press Ctrl+C to stop
──────────────────────────────────────────────────────────────────────────────────────
TYPE             │ NAME                            │    DURATION │   STATUS │ DETAILS
──────────────────────────────────────────────────────────────────────────────────────
http.server      │ GET /                           │     50.58ms │      200 │ client=127.0.0.1
http.client      │ GET api.github.com/users/github │     656.2ms │      200 │ host=api.github.com
http.server      │ GET /external                   │     662.0ms │      200 │ client=127.0.0.1
business_logic   │ custom_operation                │     100.2ms │  success │
http.server      │ GET /custom                     │     101.1ms │      200 │ client=127.0.0.1

It's basically tail -f but doesn't hurt your eyes.

What Gets Traced?

• FastAPI endpoints (automatic)
• HTTP client calls via httpx (automatic)
• Custom operations via context managers

For advanced usage, configuration options, and examples, see USAGE.md in the repository.

Comparison with OpenTelemetry

Feature	LatencyX	OpenTelemetry
Setup	1 line	50+ lines
Dependencies	Minimal	Heavy
Learning curve	Minutes	Hours/Days
Best for	Solo devs, small teams	Large enterprises

LatencyX isn't trying to replace OpenTelemetry. If you need distributed tracing across 50 microservices, use OTel. If you just want to know why your API is slow, use LatencyX.

Roadmap

Feature	Status	Notes
FastAPI instrumentation	✓ Done	Works today
HTTP client tracing (httpx)	✓ Done	Works today
Console & JSON exporters	✓ Done	Works today
Flask instrumentation	Planned	Because not everyone uses FastAPI
SQLAlchemy support	Planned	Async + sync
PostgreSQL (psycopg2, asyncpg)	Planned	Native drivers
Redis tracing	Planned	Cache tracing that doesn't lie
MySQL support	Planned	For the other half
Async jobs (Celery, RQ)	Planned	Background tasks need love too

Vote for these by opening an issue:

Feature	Why it's in maybe-land
Distributed tracing with trace IDs	Complex, needs real use cases
WebSocket tracing	Depends on demand
Sentry/Datadog integration	Only if people actually need it
Slow query detection	Might build if requested
APM tool integrations	Tell me which ones matter

If something here would make your life easier, let me know. Otherwise, it stays in maybe-land.

Contributing

This is a solo project right now, but contributions are welcome:

• Feature requests - Open an issue and tell me what you need
• Bug reports - If something breaks, let me know
• Code contributions - PRs welcome, but let's keep it simple

No formal process yet. Just open an issue or PR.

Installation

# Basic installation
pip install latencyx

# With optional dependencies
pip install latencyx[http]      # HTTP client tracing
pip install latencyx[all]       # Everything

License

MIT License - use it however you want.

---

Made for developers who want observability without the headache.