cadence

The operation, counted

Recorded, not remembered. Every figure here comes from one exported dataset — the recorded history of the source instance this operation is built in. Read it as three things at once: volume (what was built), breadth (how many systems it touched) and cadence (how steadily, and where the gaps are).

source stats/data.json · platforms Gitea + GitHub (aggregates only) · Gitea 126 repos / 9464 commits · GitHub 49 repos / 585 commits · data 2026-10-21T22:20:49Z · live data

Summary

Headline values from the current dataset window — aggregate only. Volume is the least interesting of the three: it says work happened, not that it was disciplined.

175
Repositories

24 public · 151 private

10,049
Commits

since 2023-04-07

3,202
Commits · 30d

33 repos touched

39
Languages

Three clocks, one operation

The dataset measures the outermost clock. The other two are the reason it keeps ticking.

session

One coordinator, one verdict

A session is a bounded unit of work with a supervisor and a written end: what changed, what was verified, what remains. The commits below are what survives when the session closes.

board

Every finding becomes an item

Work that is not written down as an item with a state does not exist the next day. Open items stay visible on now — including the ones that are uncomfortable to leave open.

audit

A sweep on a cadence

Read-only sweeps with a verdict per host, repeated rather than trusted once. An audit is only worth the things it is allowed to find.

Series

Volume over time, language distribution, and where the repositories sit — drawn by the server from the dataset. Read the shape, not the last point: a flat stretch is a fact, and a gap is shown as a gap.

Commits per month commits / month
Commits per month6,0004,0002,00002023-042024-032025-032025-102026-072026-10

32 months · 2023-04 → 2026-10

Language share % of language bytes
Language share
Gnuplot Rust TeX Sage Python C Shell Other (6)
Repositories by primary language repositories
Repositories by primary language6040200RustHCLLuaDJupyter NoTypeScript
Commits per week last 13 weeks
Commits per week6,0004,0002,000007-2308-1309-0309-2410-15
method

How these numbers are produced

  • Counts come from the read-only APIs of both forges, across every repository in scope; forks and mirrors are excluded so copies don't inflate volume.
  • Public and private repositories are combined into the totals — only aggregates are published, never repository names or commit messages.
  • Private is the default state of a repository here, and a visibility gate fails the publish if one is readable from outside.
  • The dataset is read by the server and rendered into SVG on the request. This page ships no analytics and sets no cookies.
  • Where a period has no recorded commits, the gap is shown as zero rather than interpolated away.
honesty notes

What this page will not show

  • A number without a window — every series states its range.
  • A guess dressed as a count: a repository whose history was too large to walk commit-by-commit is marked sampled in the dataset, and its real first and last commit dates are kept.
  • A projection. Everything here is a claim about the past; a forecast would be a claim about the future, and there is no dataset behind one.
  • Repository names, commit messages or anything private, even though the source instances hold both.
  • A dropped incident. The post-mortems are the other half of this record, and the field notes are the decisions behind it.
  • Third-party analytics, fingerprinting or session recording. There is no client-side tracking on this site at all.

Prefer the raw view? Live dashboards are on infra.