Build briefs · #295
A Zig project for the graphrag topic
No repository among the 28 repositories we track with this topic is written in Zig, while the topic gained enough stars in 7 days to clear our rising test. The numbers below are the whole argument: nothing on this page was written by a model.
Signal faded: the topic no longer clears our rising test
Topic counts and star scores as of 28 Oct (BKK).
Opened
15 Oct 2026
the first day this brief was served, written once and never changed
Window closes
13 Jan 2027
77 days left
Repositories in the topic
28
4 of them rising in the 90-day window
Written in Zig
0
of the 28 we track with this topic
The problem, measured
Counted from our own tables on every pipeline cycle. Star scores come from our daily snapshots rather than a live read of GitHub, so the numbers here are the ones the rest of the site ranks with.
| Measure | Value | Where it comes from |
|---|---|---|
| Repositories we track that carry this topic | 28 | repo_topic_history (topic seen in the last 30 days) joined to repos we track, archived and forks excluded |
| Of those, rising (4 gained the bar in 7 days) | 4 of 28 (14%) | repo_scores window 7d on the latest computed day, repositories gaining 100 stars or more (scores of 2026-10-28) |
| Stars those repositories gained in the 7-day window | +9.3k | repo_scores window 7d, sum of positive star changes in the topic (scores of 2026-10-28) |
| Repositories in this topic written in Zig | 0 of 28 | repos.primary_language, counted over the same topic set |
| Repositories we track in total (the scope of every claim here) | 63,108 | repos.is_tracked |
Who has already tried
The 20 largest repositories we can see carrying this topic, with the status each one has in our data. This list keeps repositories that are archived on GitHub, which the topic counts above exclude, so the two totals are not the same set.
| Repository | Status | Stars | Language | Last push | Open issues | Merge wait | 7-day stars |
|---|---|---|---|---|---|---|---|
| Graphify-Labs/graphify | Still developing | 137,021 | Python | 2026-10-27 | 720 | not measured | +8.7k |
| HKUDS/LightRAG | Still developing | 40,372 | Python | 2026-10-27 | 244 | 10h (371 merged) | +197 |
| microsoft/graphrag | Still developing | 36,468 | Python | 2026-10-27 | 26 | 10h (26 merged) | +98 |
| tirth8205/code-review-graph |
Why the existing ones are not enough
Every number below is counted over exactly the rows in the table above, so it can be re-derived from them. "Not measured" means we have not measured it yet, which is different from measuring zero.
| Measure | Value | Where it comes from |
|---|---|---|
| Archived on GitHub | 0 of 20 (0.0%) | repos joined to repo_topic_history for this topic, ordered by stars |
| Popular but quiet (1000+ stars, no push for 24 months) | 0 of 20 (0.0%) | repos.pushed_at compared to now() |
| Quiet and below the popularity floor | 0 of 20 (0.0%) | repos.pushed_at compared to now() |
| Still developing (pushed within 24 months) | 20 of 20 (100%) | repos.pushed_at compared to now() |
| Open issues across those attempts | 2,947 total, median 37 | repos.open_issues as GitHub reported it at the last snapshot |
| Median wait from opening a pull request to merging it | 2h (measured on 12 of 20) | repos.merge_latency_median_days and repos.merge_latency_samples |
| Longest silence in the list | 1 months | repos.pushed_at compared to now() |
The stack, and what is already there
"The same problem" is defined here as sharing repositories with this topic: the topics below are the ones that appear most often on the same repositories as graphrag. They are listed with the number of repositories that share them, because that number is the scope of the evidence.
This topic is written in, among the repositories we track: Python 13 · Rust 7 · C++ 2 · Go 2 · HTML 1. The brief proposes Zig, and this is what it would be joining.
Topics that share repositories with it: rag 20 · knowledge-graph 18 · llm 13 · mcp 11 · ai-agents 10.
Repositories written in this language, in topics that share repositories with this one, that are growing now: ddalcu/mlx-serve +134 stars in 7 days.
- ddalcu/mlx-serve · llm · 2,114 stars · +134 stars in 90 days
This is evidence that the stack is doing work in the neighbourhood of the problem, not a claim that one language is faster than another: our numbers do not support that claim (see the note at the foot of the gap map).
First milestone, and when this brief fails
Both are stated with the constants this index ranks by, so neither can drift away from the data. The milestone is the same bar that makes a repository count as rising for a topic; the failure date is exactly the window the closing check uses.
Milestone 1: a repository carrying the topic graphrag and written in Zig appears in our index and gains at least 100 stars in a 7-day window of our own snapshots. That is the same bar that makes a repository count as rising for a topic here, so it is measured with the same query as everything else on this page.
Failure: at 2027-01-13 (90 days after 2026-10-15) no repository carrying the topic graphrag and written in Zig has appeared in our index. We record that as a miss on this page and we do not delete the brief. A miss is a statement about our index only.
Closing the loop
We re-run this check on every pipeline cycle. A repository is counted only when it carries the topic and the primary language and we first saw it in our index after the brief was published, which is why the count starts at zero on the day a brief is opened.
Since 2026-10-15 no repository we track with the topic graphrag and the primary language Zig has first appeared in our index. We re-run this check on every pipeline cycle and count it here when one does.
Matching is by category: the topic and the primary language, nothing else. A repository is counted only if we first saw it after the brief was published, and the tag on its page says the same thing. There is no account, no submission and no notification in this loop: it is one query over rows we already have.
Languages offered by a brief are Rust, Go, Zig, the same three the gap map counts. Being written in one of them is not a claim about speed, safety or growth; the measure here is where work has not been done in the repositories we track. Every number on this page can be re-derived with the query named next to it, and a brief that turns out to be wrong is reported as a miss rather than removed.