What 783 graded repositories reveal

Every repo RepoGrade scores feeds these aggregates: how quality tracks reach, age, and maintenance, and what separates the best repos from the rest.

Repositories
783
Python
Median score
74
out of 100
Median stars
236
GitHub stars
Median age
1.7 yr
620 measured
Ship releases
45%
626 measured
Are forks
2%
620 measured

How the corpus scores

Each dot is a repo, colored by grade. Corpus medians divide the four quadrants. Distributions below show the same data sliced by band and popularity tier.

Quality vs popularity

Hover a region to highlight it and see counts. Toggle the X axis between stars and contributors.

X axis
median 236med. 74HIDDEN GEMS122 · 17%EXEMPLARY279 · 39%NEEDS WORK206 · 29%VIRAL, MESSY113 · 16%02550751001101001k10k100kStars (log scale)Score

Score distribution

Repos per 10-point band, colored by grade

Score by popularity tier

Box = P25-P75, line = median, whisker = P10-P90. Colored by grade.

Group by
0204060708090100ABCDF< 10n=5010-100n=128100-1kn=3101k-10kn=12610k+n=106

GitHub stars · box = P25-P75, line = median, whisker = P10-P90

Scores climb steadily with popularity

Median score within each star tier. The grade mix hides this because most repos have few stars - but quality and reach rise together.

What top repos do that the rest don't

Average pass rate for A/B vs D/F repos, by dimension and by check. Filtered to Python. Based on 530 repos.

Quality gap, by dimension

Average check pass rate across all checks in each category, for top-graded vs. bottom-graded repos.

Documentation+46pp
A/B
88%
D/F
42%

4 checks averaged

Engineering+61pp
A/B
88%
D/F
27%

6 checks averaged

Project health+27pp
A/B
98%
D/F
72%

2 checks averaged

Pass rates side by side

Each row is one check. Filled = A/B pass rate, hollow = D/F. +Npp = percentage-point gap.

D/F passA/B pass0%25%50%75%100%CI/CD+83ppTests+74ppLinting and formatting+70ppContributing guide+60ppIssue and PR templates+57ppInstall and run instructions+50ppReproducibility+42ppDependency manifest+40ppREADME+37ppLicense+37ppHousekeeping+32ppRepository metadata+21pp

All checks

CheckA / B pass rateD / F pass rateGap
CI/CD
90%
7%
+83
Tests
99%
25%
+74
Linting and formatting
80%
10%
+70
Contributing guide
67%
7%
+60
Issue and PR templates
68%
11%
+57
Install and run instructions
92%
42%
+50
Reproducibility
93%
51%
+42
Dependency manifest
97%
57%
+40
README
95%
58%
+37
License
97%
60%
+37
Housekeeping
98%
66%
+32
Repository metadata
98%
77%
+21

Does a better grade attract more attention?

Median traction signals per grade band. Toggle across stars, forks, watchers, open issues, and contributors.

Median traction by grade

Toggle across stars, forks, watchers, issues, contributors

Metric

How actively are repos maintained?

Release discipline and project posture signals - the upkeep story behind a grade.

Release discipline by grade

Share of repos shipping at least one tagged release, per grade band

Project posture

Fork quality penalty, archived repos, and release cadence

Quality: forks vs. original repos

Originals
67
Forks
31

Forks score 36 points lower on average (11 forks of 620 measured)

Archived (read-only)0%

0 of 620 measured

Ship releases45%

284 of 626 measured

Does age or size move the score?

Median score across how long a repo has existed and how large it is on disk.

Score by repo age

Median score across age cohorts

Score by repo size

Median score across size on disk

Most common topics, by quality

Topics repos tag themselves with - ranked by frequency, with the median score of all repos carrying each tag.

Top 20 topics
1python
79
84
2llm
40
80
3claude-code
27
73
4ai
26
84
5ai-agents
25
88
6vless
17
44
7v2ray
17
44
8mcp
17
86
9proxy
16
46
10claude
16
80
11agent
14
82
12codex
13
83
13vmess
13
46
14vpn
12
46
15shadowsocks
12
46
16multi-agent
11
82
17hacktoberfest
11
77
18machine-learning
11
80
19open-source
10
84
20fastapi
10
86