Blog
Engineering
Engineering
How AI Coding Agents Find Context: 7 Codebases Read
How do AI coding agents find context? We read the code of 7 open-source agents: none indexes your repo by default. They grep, read, then trim the window.
Read11 min read
Engineering
Inside the 400-commit-a-day repos
openclaw and Hermes Agent each land 400 to 700 commits a day. We measured the pace, the authors and the code shape, and set it against the published research.
Read8 min read
Engineering
Deleted per 100 added: mature projects delete, young ones mostly add
requests has deleted 82 lines for every 100 it added in 15 years. Projects started since 2025 delete about 42. What the ratio says about how code ages.
Read7 min read
Engineering
The 6,967-line function: the longest functions in popular open-source repos
The longest function in a 10k+ star repo is 6,967 lines. We hand-checked the top ten at the commit we indexed and sorted why functions get that long.
Read8 min read
Engineering
The Bus Factor of 35 Popular Open-Source Repos
In 13 of 35 popular repos, one person is the main author of over half the files. Hugo and FastAPI each have 900+ commit authors and one main author.
Read6 min read
Engineering
Prefect vs Airflow vs Temporal: three ways to decide what runs next
Airflow locks database rows in a scheduler loop, Prefect runs ordered rules on every state change, Temporal replays an event log. Read in the source.
Read9 min read
Engineering
Python Web Frameworks Compared: What Is Inside Seven Repos
FastAPI, Django, DRF, Flask, Litestar, Robyn and Sanic read from source: where each routes and validates a request, with a health score per repo.
Read10 min read
Engineering
Rust UI three ways: how ratatui, iced and bevy draw a frame
ratatui vs iced vs bevy: how each Rust UI library draws a frame and where it keeps state, traced through the real source, with code health scores.
Read9 min read
Engineering
GitHub Stars vs Code Quality: We Scored 3,126 Open-Source Repos
Median code health falls from 8.50 for repos under 100 stars to 6.80 at 10k+, but compare repos of the same size and stars barely move the score.
Read7 min read
Engineering
Vibe coding and code health: what six studies say, and what 3,126 repos show
Fast, AI-heavy repos score lower on code health at the same size, while reviewed agent commits are no buggier than human ones. Six studies and our data.
Read7 min read
Engineering
Where architecture decisions actually live: 190 popular repos, 10 with ADR files
We mined 10,339 decisions from 190 popular open-source repos. 154 record them in pull requests, 163 in commit messages, and at least 10 in ADR files.
Read8 min read
Engineering
35x or 16%? We Published Both Token Benchmarks
One tool, one codebase: 35.6x fewer tokens for a single payload, 15.9% fewer in an agent loop. Why both are correct, and what each benchmark measures.
Read7 min read
Index your repo free.
See what repowise finds in your own code in minutes. Or explore a live repo first.