lightlogger

Live web dashboard for your Python logs — pip install, add one line, open localhost:4356. Zero dependencies.

Install

pip install lightlogger

import lightlogger
lightlogger.start()
lightlogger.info("user logged in")

That's it — open http://127.0.0.1:4356 and watch it stream in live, in a dark, searchable dashboard. No config file, no separate server process to run, no npm install.

Your existing logging calls show up too, with zero code changes:

lightlogger.start(capture_logging=True)  # the default
import logging
logging.getLogger("some.third.party.lib").warning("disk almost full")  # appears in the UI automatically

And when one log message is really four related operations, group them:

with lightlogger.group("process_order #4821"):
    lightlogger.info("validating cart")
    lightlogger.info("charging payment", data={"amount": 49.99})
    with lightlogger.group("send_notifications"):
        lightlogger.info("email sent")
        lightlogger.info("sms sent")

They render as a collapsible tree — click to expand, nested groups indent, a red badge appears if anything inside failed.

Features

API reference

start() parameters:

Built-in help

Forgot the API? It's in the package, not just this README:

>>> import lightlogger
>>> lightlogger.help()

Prints a full cheatsheet to your terminal — every function, one-line descriptions, tiny examples, the capture_logging gotcha, no need to leave your shell. And once the dashboard is running, open http://127.0.0.1:4356/help for the same reference as a page, with copyable code blocks.

Why lightlogger

Debugging a running Python process usually means one of three things: print() statements you'll forget to remove, a terminal window full of scrolling text you can't search, or reaching for a heavyweight observability platform to answer a question that takes ten seconds to answer once you can actually see your logs.

We looked at what else exists. Logdy is a Go binary you install separately from your app. Chronologer needs its own server process. cutelog needs PyQt. Logfire is closed-source and cloud-hosted. lnav and klp are terminal-only. Django and Flask debug toolbars only work inside those specific frameworks, in that specific request/response cycle.

None of them do the one thing that actually matches how a Python developer debugs: import, call one function, and get a live web page — from inside the process you're already running, with no extra install, no extra service, no dependencies to audit. That's the gap lightlogger fills. And once your logs have a real UI instead of a scrolling terminal, grouping related operations into a collapsible tree stops being a nice-to-have — it's the difference between reading a wall of text and reading a story.

Security

lightlogger binds to 127.0.0.1 only, by default — nothing is reachable outside your machine unless you explicitly pass host="0.0.0.0" (which prints a loud warning when you do). It's a local development tool, not a production observability system: don't run it against a public-facing process, and don't leave it running longer than your debugging session needs.

Contributing

Contributions are welcome — see CONTRIBUTING.md for the ground rules (short version: zero dependencies, stay on stdlib, keep it simple). Bug reports and PRs both go through GitHub Issues.

License

MIT — see LICENSE.