dshseek

seek://news/upstream-200k

DeepSeek Harness Passes 200,000 Stars

Sep 2, 2026ecosystem#deepseek-harness #milestone #stats← All news

TL;DR

Eighteen days after this site's 100,000-star note, the upstream repository has passed 200,000 GitHub stars — roughly a hundred thousand more than the just-over-108,000 reading that note cited. Nine releases have been published in the meantime, every one flagged prerelease. As always on DSHSeek, the number comes from the build-time snapshot pipeline, not from anyone's hands.

The counter on the stats page has crossed another class boundary: DeepSeek Harness has passed 200,000 GitHub stars. Today’s API fetch reads past 208,000.

Timing the crossing

The 100,000-star note ran on August 15, when the build-time snapshot read just over 108,000. The committed snapshot from August 30 already reads 203,774 — so the 200,000 mark fell inside those fifteen days. Between August 30 and today, roughly five thousand more stars arrived. Eighteen days, roughly a hundred thousand stars: the vertical phase the August note described never leveled off.

What arrived with the stars

One thing the 100,000-star note flagged has genuinely changed: the repository had no releases then, and it now has nine, published between August 17 and September 2. Every one is flagged prerelease — four of them landed on four consecutive days, and the notes themselves keep warning about compatibility breaks. Velocity is real; stability is still a roadmap item.

The star curve is not happening in a vacuum. The community desktop client on this site’s map, DSH Desktop, has itself passed 22,000 stars; plugin routers, sidebar workbenches and memory plugins each draw their own audiences. The ecosystem survey described a catalog taking shape — at twice the star count, the shape is holding.

What it still does not mean

The same two caveats as August, still true: not stable, and not a substitute for judgment. The README says it outright — developer preview, iterating rapidly, compatibility-breaking changes announced. Teams adopting at 200,000 stars face the same trade-off as teams adopting at 100,000: pin versions, expect churn.

How we count

Same methodology as always, because this site’s premise is verifiability: star counts on DSHSeek are never typed by hand. The build fetches them from the GitHub API into a committed snapshot refreshed daily at 03:00 UTC, and every page renders from that snapshot. When this number disagrees with the badge on GitHub, the build is stale, not the claim.

Official references