workloadworth a look
24 nights with commits after midnight in the last 90 days
11 in the 90 days before
gitdashx 0.3.0 · windows x64 · runs offline
See how your team really builds software.
Point it at your GitHub organizations and it reads the commit history the way a colleague would: when the work actually happened, what shipped each quarter, how pull requests moved, who is the only person who knows a product.
It is a desktop app. There is no account, no server of ours, and nothing to upload.
125.8 MB · windows 10/11 · no node, no git, no gh to install first
1,271 commits · 14 repositories · 6 people · last 12 monthslessmore
#preview
This is the whole thing
One window, every organization you belong to, for any period you choose.
gitdashx / localyou
commits
1,271
+14%
merged prs
235
+8%
after hours
18%
was 11%
median to merge
6h40m
-1h20m
esapghgauditportal
contributors
| a.rahman | 412 | 24% | |
|---|---|---|---|
| s.kowalski | 318 | 9% | |
| m.okafor | 247 | 31% | |
| j.lindqvist | 174 | 6% | |
| t.nakamura | 120 | 12% |
pull requests
- opened
- 268
- merged
- 235
- within a day
- 72%
- waiting longest
- 19 d
Sample data. A screenshot of a real run would show real people's working hours.
#charts
Sixteen charts, one window
Everything the app draws, at the size it draws it. Pick a person and a period and all of it follows.
your share of team output
how much of the team's commits, pull requests, reviews and issues are yours
commits
all branches
41232%
prs opened
open + closed
9636%
prs merged
authored, merged
8837%
prs reviewed
others' prs
6133%
issues opened
open + closed
4435%
leaderboard
everyone, all time, every count side by side
| contributor | commits | prs | merged | rate | reviewed | issues | repos |
|---|---|---|---|---|---|---|---|
| a.rahmanyou | 412 | 96 | 88 | 92% | 61 | 44 | 9 |
| s.kowalski | 318 | 71 | 64 | 90% | 52 | 31 | 7 |
| m.okafor | 247 | 54 | 47 | 87% | 38 | 26 | 6 |
| j.lindqvist | 174 | 32 | 28 | 88% | 24 | 18 | 5 |
| t.nakamura | 120 | 15 | 8 | 53% | 9 | 7 | 3 |
highlights
up to six ranked observations, each set against the period before
workload
49% of your commits in the last 90 days were outside office hours
down from 53% - rest of team 1%
workloadworth a look
15 days in a row with commits, without a day off
28 jul - 11 aug, the longest stretch this year
output
+110% commits in the last 90 days against the 90 before
544 against 259
focusbetter
6h 40m median time to merge, down from 8 hours
72% merged within a day
team riskworth a look
5 products where one person wrote over 80% of the commits
esap 92%, ghg 85%, audit 100%, and 2 more
when the commits happen
after hours, late nights, weekends, busiest hour
outside office hours
18%
229 of 1,271 commits
rest of team 7.7%
late night, 22:00-04:59
11%
140 commits
rest of team 1.4%
after midnight
64
on 41 different nights
rest of team 5
weekends
13%
162 commits, 30 weekend days
rest of team 1.1%
busiest hour
12 pm
127 commits in that hour
rest of team 5 pm
hour x weekday
office hours boxed, the late-night bands marked
share of commits by hour
your bars, the rest of the team as a line
by weekday
the team's share sits on the same track as a tick
mon23.3% · 297
tue16.9% · 215
wed15.2% · 193
thu14.3% · 182
fri17.6% · 224
sat7.5% · 96
sun5.2% · 66
every day, every commit
a year at a glance, with the streaks and the gaps underneath
days with a commit
214
of 365
longest streak
15 days
28 jul - 11 aug
longest gap
11 days
over the summer break
busiest day
23 commits
14 may
what was built each quarter
stacked by product, with repositories grouped into the product they belong to
esapghgauditportal
quarter x contributor
the same table also does products and repositories
rowsproductrepositorycontributor
| q2 24 | q3 24 | q4 24 | q1 25 | q2 25 | q3 25 | q4 25 | q1 26 | q2 26 | total | |
|---|---|---|---|---|---|---|---|---|---|---|
| a.rahman | 10 | 12 | 17 | 32 | 40 | 49 | 30 | 20 | 38 | 248 |
| s.kowalski | · | 12 | 15 | 33 | 46 | 44 | 50 | 42 | 29 | 271 |
| m.okafor | · | · | 29 | 36 | 42 | 29 | 30 | 41 | 29 | 236 |
| j.lindqvist | · | · | · | 38 | 35 | 45 | 23 | 23 | 27 | 191 |
| t.nakamura | · | · | · | · | 54 | 28 | 34 | 34 | 22 | 172 |
| total | 10 | 24 | 61 | 139 | 217 | 195 | 167 | 160 | 145 | 1118 |
code frequency
lines added and removed, with generated monsters clipped and marked
+500,289added-286,704removed+213,585net · weekly
pull requests
opened, merged, and how many landed over time
opened
268
merged
235
median to merge
6h 40m
within a day
72%
time to merge
your distribution against the rest of the team's
under 1 hour18%
1-4 hours27%
4-24 hours27%
1-3 days16%
3-7 days8%
over a week4%
waiting longest
open, not draft
| pull request | author | open for |
|---|---|---|
esap#412 Split the reporting job off the request thread | m.okafor | 19 d |
ghg#118 Scope 3 factors for the 2026 dataset | s.kowalski | 12 d |
portal#77 Keep the filter state in the URL | a.rahman | 6 d |
audit#204 Retry the evidence upload on a 502 | j.lindqvist | 2 d |
after hours, by contributor
who is carrying the evenings and the weekends
| contributor | commits | outside | late night | midnight | weekend | busiest |
|---|---|---|---|---|---|---|
| a.rahman | 412 | 24% | 11% | 64 | 13% | 12 pm |
| s.kowalski | 318 | 9% | 2% | 6 | 3% | 10 am |
| m.okafor | 247 | 31% | 14% | 41 | 18% | 11 pm |
| j.lindqvist | 174 | 6% | 1% | 2 | 2% | 2 pm |
| t.nakamura | 120 | 12% | 4% | 9 | 5% | 4 pm |
contributors
a card each, with a year of weekly commits
a.rahman#1
Amina Rahman
412 commits+214,880-96,412
s.kowalski#2
Stefan Kowalski
318 commits+151,204-70,338
m.okafor#3
Ada Okafor
247 commits+132,907-88,140
j.lindqvist#4
Jo Lindqvist
174 commits+61,442-22,015
t.nakamura#5
Rin Nakamura
120 commits+38,119-11,760
Drawn, not screenshotted, and every number invented. A screenshot of a real run would show real people's logins and the hours they work.
#measures
Six views of the same history
Every organization you belong to, separately or combined, for any period you choose.
- highlights
- Up to six ranked observations for the person and period you pick, each set against the period before: after-hours climbing, weekends worked, the longest run without a day off, products only one person knows.
- when-commits-happen
- An hour-by-weekday grid, hourly and weekday charts, and a calendar of every single day. Office hours adjust in 15-minute steps; everyone can be converted to one timezone.
- quarterly-delivery
- A stacked chart by product and a quarter-by-product table, with repositories grouped into the products they belong to instead of listed as raw repo names.
- code-frequency
- Lines added and removed over time, and a card per contributor, modelled on GitHub's own graphs. Vendored and generated monsters are cut off and marked, not allowed to flatten the chart.
- pull-requests
- Opened and merged, median time to merge, share merged within a day, the distribution, and the pull requests that have been waiting longest. Drafts do not count as waiting.
- team-comparison
- Share of commits, pull requests, merges, reviews and issues, a leaderboard, and after-hours numbers per person - with two accounts belonging to one person merged.
#output
It tells you what changed
Not a wall of totals. Ranked observations, each one compared with the period before it.
| after-hours share | 18% | +7 pts |
|---|---|---|
| weekends worked | 6 d | +2 d |
| longest run, no day off | 34 d | +11 d |
| late nights, 22:00-04:59 | 12 | -3 |
| median time to merge | 6h40m | -1h20m |
| products only you know | 3 | unchanged |
Each highlight is measured against the period before it, so the page tells you what changed rather than what is merely large.
#private
Where the data goes: nowhere
The whole tool is one process on your machine talking to GitHub. That is the entire architecture.
- network
- api.github.com only, with your own token
- server
- none - the app runs one on 127.0.0.1 for itself
- account
- none to create
- telemetry
- none
- source code
- never downloaded: git clone --mirror --filter=tree:0 fetches commit metadata and skips file contents
- token
- encrypted with the OS keystore (DPAPI on Windows), held by the app process, never handed to the page
- listener
- refuses requests whose Host is not localhost, cross-site requests by Origin, and non-JSON writes
- your data
- a folder on your machine; delete it and nothing of it survives elsewhere
Contribution analytics means reading how your colleagues spend their working hours. That is exactly the kind of data that should not be sitting on somebody else's server.
#how
From download to dashboard
- 01installOne .exe. Nothing else to set up: no Node, no git, no GitHub CLI - the app carries what it needs.
- 02connectA short code approved in your browser, a personal access token, or an account the GitHub CLI already holds.
- 03selectTick the organizations you want, optionally your personal repositories, and press Analyze.
- 04leave itThe first sync pulls history; later ones fetch only what is new, hourly, in the background while you read.
#download
Download
Free. Unsigned, so verify the checksum if you would rather check than trust.
| build | file | size | notes |
|---|---|---|---|
| installer | ↓ gitdashx-0.3.0-x64.exe sha256 812489d26186d26d0d6a07a30e70f4cc30d050429dde531572ad35d1642a3933 | 125.8 MB | shortcuts, per-user, no admin rights |
| portable | ↓ gitdashx-0.3.0-portable.exe sha256 5ec4f6a480ca582cf727381fe9c1cc03206d635bc3b927c6d5ab8df6c1b0b10a | 125.6 MB | runs from anywhere; one copy at a time |
- requires
- Windows 10 or 11, 64-bit. Nothing else.
- first run
- These builds are not code-signed, so SmartScreen shows Windows protected your PC. Choose More info, then Run anyway - or check the SHA-256 above against the file you downloaded first.
#faq
Questions
Does anything leave my computer?
Only requests to GitHub's own API, made with your token. There is no GitDashx server, no account and no telemetry. Everything it works out is stored in a folder on your machine.
Does it download our source code?
No. It mirrors history with git clone --mirror --filter=tree:0, which fetches commit metadata and skips file contents entirely. It never reads what is inside your files.
What access does it ask for?
A GitHub token with the repo and read:org scopes. It reads commit metadata, pull requests and repository statistics, and never writes to your repositories.
Why does Windows warn me when I run it?
The builds are not code-signed, so SmartScreen flags an unknown publisher. Choose More info, then Run anyway. Every release publishes a SHA-256 checksum you can verify first.
Do I need git or the GitHub CLI installed?
No. The desktop app carries its own copy of git. If the GitHub CLI happens to be installed and signed in, the app offers those accounts as a shortcut, but it does not need them.
Is there a macOS or Linux build?
Not yet. The packaged app is Windows x64 only today. The tool underneath is plain Node.js and runs anywhere from a checkout.