Logier · desktop log viewer
The answer is already in the file.
Logier reads the whole thing on your machine, works out its format, collapses it to the handful of message shapes it actually contains, and marks the ones that broke.
$39.99 once — not a subscription.
You already know the shape of this.
Something broke at 07:07. You have the file open, and you press Cmd‑F for error. Four thousand matches. Thirty‑nine hundred of them are the same retry loop firing every two hundred milliseconds, which you’ll work out somewhere around match six hundred.
The line that explains the outage is in there once, and it doesn’t contain the word error.
parsed
rawEvery log is a format.
Logier works out which.
Eleven parsers ship in the box. Detection walks them in order and commits each line to the first that claims it, so nothing has to be chosen before a file will open. A fallback parser sits behind them and never refuses a line.
When none of them fit — the normal case for a team’s own services — you describe the format once and Logier drafts the pattern from the file rather than handing you a blank box. The result is plain JSON, so it can live next to the service that writes the log.
- Android
- Apache
- OpenSSH
- Proxifier
- Spark
- Zookeeper
- Thunderbird
- Windows CBS
- BGL
- Linux syslog
- macOS
- + your own, as a schema
one line, as it arrives
Dec 10 06:55:48 LabSZ sshd[24200]: Failed password for invalid user webmaster from 173.234.31.186 port 38926 ssh2timestamp
Dec 10 06:55:48
host
LabSZ
process
sshd[24200]
levelderived
ERRORmessage
Failed password for invalid user webmaster from 173.234.31.186 port 38926 ssh2
Most of a log file is the same line.
Logier reduces every message to its shape, masking the parts that make each occurrence unique. Clustering and anomaly detection share that one definition of “the same message” — which is why a repeat storm is simply a cluster that grew large enough to be worth telling you about.
Five ways a file admits
something went wrong.
Findings are ranked by severity and capped, so the panel stays a list you can read rather than a feed you scroll. Each one carries the numbers it was derived from — the rate it moved from, the interval it broke, the share of the file it now accounts for.
entries
silence gap · high
sshd stopped logging entirely between 07:03 and 07:07, against a typical interval of under two seconds.
frequency spike · high
ERROR rate jumped from ~1/min to 18/min on resume and held there for eight minutes.
- 01
Frequency spike
A level whose rate jumps against its own trailing baseline and stays there. The finding carries the rate it moved from, the rate it moved to, and how long it held.
- 02
New signature
A message shape that has not appeared before in this file. If it goes on to repeat, the finding says how many more times it occurred.
- 03
Silence gap
A service that stopped logging entirely. The absence of lines is the finding — measured against that service's own typical interval, not a fixed threshold.
- 04
Outlier value
A duration in the message text that runs long against what is typical for its neighbours. One entry took 4.2s where the usual is 30ms.
- 05
Repeat storm
One message shape grown large enough to account for a real share of every error and warning in the file — the cluster that became worth reporting.
The model gets an excerpt. Never the file.
Copilot is bring‑your‑own‑key — OpenAI, Anthropic, Google, or any OpenAI‑compatible endpoint, including a model running on your own machine. Parsing, clustering, the timeline and every finding are computed locally whether or not it is configured at all.
What gets sent is the context around the thing you asked about, scaled to your token budget, and the app tells you exactly what it included — so the answer’s blind spots are visible rather than implied.
In a generated report, the figures render from the computed payload, never from model text. The prose can be written for you. The numbers can’t.
Why did this spike start at 07:07?
A.sshd stopped logging for four minutes beforehand, then resumed at roughly nine times its earlier rate. Every entry since shares one signature — a failed password for an invalid user — from 41 distinct addresses.
Nothing leaves this machine unless you ask it to.
- Files are parsed in memory, on your machine. Opening one needs no account and no upload.
- Cloud storage, if you turn it on, is your own Dropbox, Google Drive or OneDrive. Logier's servers never hold your files.
- Settings sync is manual and names the fields it sends. API keys are not in the payload and are never read out of one.
- A theme imported from the web builder is treated as hostile input and rebuilt field by field before anything touches disk.
why this exists
Why isn’t there a tool that can make sense of any log format I run into?
— Paul Pietzko, creator of Logier
At my company we run an in‑house CMS called 2sxc. Whenever I updated a client’s site and something went wrong, I found myself digging through plain‑text log files, searching for the word ‘error’ and trying to piece together what had happened. That question is the whole reason Logier exists.
The long version
or see them on the app →- 01
Auto-format detection
Apache, syslog, SSH, Android, Spark, Zookeeper, and more — Logier identifies your format automatically, with a fallback parser for anything else.
- 02
Live Mode
Stream a remote log endpoint and watch new lines land in real time. Filter and search without missing a beat.
- 03
AI root cause
Copilot reads the context around an error and writes a plain-English explanation of what broke, when, and why.
- 04
Timeline view
Visualise error spikes over time. Pinpoint exactly when things started going wrong at a glance.
- 05
Anomaly detection
Sudden frequency changes, outlier values, and unusual patterns.
- 06
Secure by default
Logs processed in-memory, never leaving your system without consent.
Open the file. Get the answer. Close it.
$39.99, once. Not a subscription, not a seat, not a usage tier that grows with how useful the tool turns out to be.