Please check out my fist bigger open source project. Logic
Recently I was thinking what project I should work on and during that I remembered that during my professional career as a full stack developer I spent a lot of time investigating problems, debugging them. I had to check a lot of long files to find the source of the problems. I tried to use many different log analysis application but neither of them were to my liking so I started to develop my own log analysis application, Logic.
During the development one of the mist helpful tool was the Claude design. For me it's really hard to come up with UI designs and the manual result just didn't feet right to me. With this tool I could give my ideas, requirements, design and color themes, layouts and it gave me multiple good looking options. After I selected my favorite design, it even created the skeleton code for my app. And with this I could concentrate on the core of the app.
I hope some of you will try it and it will get your liking.
It's still under development. Feel free to join in it.
markdamics
/
Logic
Log analyzer
Logic
Log Analyzer — a full-stack app for registering log sources and inspecting their content in real time.
Core features: source management, real log ingestion with filtering/sorting/pagination, a live-updating Log Stream, a Dashboard overview, a durably indexed Search & Query bar (Lucene syntax plus SPL and LogQL subsets, with count/rate/numeric-stat aggregation charts), threshold and anomaly Alerting with webhook notifications, log-native observability (trace-ID correlation across sources and an optional APM deep-link), and Saved Searches — styled after the Axiom HUD design system, with four selectable themes.
Features
Log Stream
-
Real ingestion: tails the trailing window of each source (bounded reads — 512KB / 3,000 lines per file) rather than loading it whole, with level/timestamp/stack-trace-continuation parsing that recognizes this app's own log format, ISO-8601, Apache/Combined access logs, and BSD/RFC-3164 syslog (falling back to keyword-based severity — "failed" "denied", "exited on signal", etc. — when a line carries no explicit level tag, as…
Top comments (3)
Congrats on shipping it - the reason log tools are hard is that everyone's muscle memory is
grepplus scrolling, so the bar to open something else is high, and "none of them were to my liking" is the honest reason to build one.The question I'd want answered for my own workflow is what the unit of analysis is. Most analyzers group by request/trace id, which only works if your logs already carry one, and the interesting cases are usually the ones where they don't. If Logic can cluster similar lines and collapse repeats, that's the feature that would actually replace my grep loop - is the grouping done in-app, and does it survive a service that changes its log format between deploys?
Thank you for questions. There is no feature yet for your use case but I had some vague ideas before to group the logs which are after the other. After your comment I looked into it more and I have and idea for this but it's a bigger feature. But because of your interest in my app I am gonna implement this as next feature.
I plan to implement a new service which would be a template miner. I want to tokanize each message and within a threshold match with other messages
Thank you again for you comment and interest.
I added a new feature called patterns. All the patterns are mined from the logs by tokenization. Please check it out, I hope it will replace your grep loop. I have a roadmap in the docs folder, you check the features I want to implement in the future.