Skip to main content

Zero-config observability for Python web frameworks

Project description

  Brakit Python SDK

AI writes your code. Brakit watches what it does.
Every request, query, and API call mapped to the action that triggered it. See your entire backend at a glance.
Open source · Local first · Zero config · AI-native via MCP

GitHubWebsiteContributing

MIT License PyPI version Python versions npm version PRs welcome!


Brakit Actions — endpoints grouped by user action
Every endpoint grouped by the action that triggered it — see "Sign Up" and "Load Dashboard", not 47 raw requests

Brakit Dashboard — issues surfaced automatically
Brakit catches N+1 queries, PII leaks, and slow endpoints as you develop

Claude reading Brakit findings in VS Code
Claude reads issues via MCP and fixes your code

Quick Start

The Python SDK captures queries, fetches, and requests from your FastAPI or Flask backend and forwards them to the Brakit dashboard running in your Node.js frontend.

1. Install the Node.js core (runs the dashboard)

npx brakit install

2. Install the Python SDK

pip install brakit

3. Add one import to your app

FastAPI:

import brakit  # must be before FastAPI import
from fastapi import FastAPI

app = FastAPI()

Flask:

import brakit  # must be before Flask import
from flask import Flask

app = Flask(__name__)

That's it. Start both apps normally. The Python SDK auto-detects your framework, instruments queries and fetches, and forwards everything to the Brakit dashboard at http://localhost:<port>/__brakit.

Full setup guide: brakit.ai/docs/introduction


What You Get

  • Cross-service tracing — FETCH from Next.js to FastAPI? The DB query inside FastAPI nests under the fetch automatically
  • Full server tracing — DB queries, outgoing HTTP calls, errors — zero code changes
  • Live dashboard at /__brakit — performance overview, request timeline, scatter charts
  • 8 security rules scanned against live traffic — leaked secrets, PII in responses, missing auth flags
  • N+1 query detection — same query pattern repeated 5+ times in a single request
  • AI-native via MCP — Claude Code and Cursor can query issues and verify fixes directly

Supported Frameworks

Framework Status
FastAPI Supported
Flask Supported
Django Planned

Supported Databases

Driver Status
SQLAlchemy Supported
asyncpg Supported
psycopg Planned

Supported HTTP Clients

Library Status
urllib3 Supported
httpx Supported
aiohttp Supported
requests Supported (via urllib3)

How It Works

import brakit  →  auto-detect framework  →  instrument everything
                        |
                        +-- DB queries    (SQLAlchemy, asyncpg)
                        +-- HTTP fetches  (httpx, aiohttp, urllib3)
                        +-- Errors        (global exception hooks)
                        +-- Logs          (logging module)
                              |
                              +-- Forward to Node.js dashboard via localhost

import brakit patches your framework's middleware to capture request/response pairs, hooks into database drivers and HTTP clients, and forwards telemetry to the Brakit dashboard running in your Node.js process. No proxy, no extra ports.

Production Safety

Brakit never runs in production. Multiple layers ensure it:

Layer How it blocks
should_activate() Checks NODE_ENV + 15 cloud/CI env vars
devDependency Not installed in production
safe_wrap + circuit breaker Errors in brakit code fall back silently
BRAKIT_DISABLE=true Manual kill switch

Optional Dependencies

# FastAPI support
pip install brakit[fastapi]

# Flask support
pip install brakit[flask]

# Development (tests, linting)
pip install brakit[dev]

Development

git clone https://github.com/brakit-ai/brakit.git
cd brakit/sdks/python

pip install -e ".[dev]"
pytest tests/ -v

Contributing

Brakit is early and moving fast. Areas where help would be great:

  • Database adapters — psycopg, Tortoise ORM, Django ORM
  • Framework adapters — Django, Starlette
  • HTTP client hooks — more client libraries
  • Security rules — new patterns, configurable severity

Please open an issue first for larger changes so we can discuss the approach.

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

brakit-0.1.6.tar.gz (31.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

brakit-0.1.6-py3-none-any.whl (47.0 kB view details)

Uploaded Python 3

File details

Details for the file brakit-0.1.6.tar.gz.

File metadata

  • Download URL: brakit-0.1.6.tar.gz
  • Upload date:
  • Size: 31.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.9.6

File hashes

Hashes for brakit-0.1.6.tar.gz
Algorithm Hash digest
SHA256 e80fb42f72177c5c4dbaddce0ff4ed14b1c3e4b9e18088497762fd70c715073f
MD5 26ed4664deb685704a62126f7e6b1c8c
BLAKE2b-256 36c551256b055b04b8543f05509ea69513060aae93ef7b7f223bfeba790144a5

See more details on using hashes here.

File details

Details for the file brakit-0.1.6-py3-none-any.whl.

File metadata

  • Download URL: brakit-0.1.6-py3-none-any.whl
  • Upload date:
  • Size: 47.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.9.6

File hashes

Hashes for brakit-0.1.6-py3-none-any.whl
Algorithm Hash digest
SHA256 e774e22df7ac68d8acf592e7119c79e2c2be618428a62836851dfb15e0e8c4a9
MD5 360570d74dd8428a5d33ca3d07389b4d
BLAKE2b-256 54293fb1cdaf62954f815802b11338085eb70d1f4df056667fd582a12f0e376f

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page