Every figure on this site is
reproducible from a public API call. This page states exactly what was called, what was
computed, and what the result cannot tell you.
What this is, and what it is not
A point-in-time comparison of GitHub adoption signals
across a fixed set of AI coding tools, read the same way on the same day. It is not a
measure of quality, usage, revenue or developer satisfaction, and nothing here should be read
as a recommendation.
Cadenceone harvest per scheduled run; each harvest is one reading
Entities41 repositories,
fixed list, reviewed by hand
A failed repository read is recorded as a
gap and produces no row. It is never written as zero and never interpolated, because a
missing measurement and a measurement of nothing are different facts.
Periods
A period is a calendar month. Each harvest writes its own
timestamped snapshot, so a period can hold many readings.
The figure published for a period is its last reading. While the period is the newest one
it is open: it gains readings and its figure can change. Once a later period exists it is
closed and never changes again. Every provenance page states which of the two it is, so a
figure can be trusted to stay put only when it says it will.
Derived figures
Two of the five figures are computed, not read. Both are
reproducible from the other columns in the same export, so they can be checked rather than
believed:
Stars Per Daystargazers_count / max(days since created_at, 1)
Fork-to-Star Ratioforks_count / stargazers_count * 100, or 0 when
stars are 0
These repositories were requested and did not resolve.
They have no row in this period:
None in the current period.
Limitations
No usage datastars and forks are not downloads or active installs
No revenue or pricingthis series collects none yet
No quality signalcode quality, security and maintenance are not measured
Fixed entity listtools absent from the list are not evidence of absence
GitHub onlyclosed-source tools have no repository and cannot appear
Corrections
If a reading is wrong, a later harvest corrects the
open period, and the provenance page carries the new value with its reading time. A
closed period is never rewritten: a number that has been cited must keep meaning what it
meant. If a correction is material, it is noted in the repository's commit history, which is
public.