Pulse
2026-09-18-v3.0
Dataset
2026-10-27-catalog-ai-v2
Ruleset
b2e8889c66f7
Scanned
2026-10-27
Window
February–September 15, 2026
Catalog tracking observation. This repository is measured with the same fixed window and ruleset, but it is not part of the published benchmark or its aggregate claims.

UCCL is an efficient communication library for GPUs, covering collectives, P2P (e.g., KV cache transfer, RL weight transfer), and EP (e.g., GPU-driven)

uccl-project/uccl ↗

uccl-project/uccl AI coding statistics: 6.5%

of 260 commits in February–September 15, 2026 were committed by an AI agent — a floor, not an estimate
6.5% agent · top contributor 11.8%11 contributors
org-wide

Adoption is broad. Many contributors are committing agent-authored work, which is what a real organisational rollout looks like.

17
agent commits
243
other considered commits
11
agent contributors
11.8%
largest contributor

What shapes this result

Mode, timing, and detected signatures show whether the result reflects broad adoption or one concentrated workflow.

Assisted or autonomous

17assisted
0autonomous

Assisted — a developer’s identity authored the commit and the agent added itself as co-author. Autonomous — a bot identity authored the commit without a human identity attached.

Agent share by month

Agent-attributed commit share by month
MonthAgent shareConsidered commits
2026-020%59
2026-033.1%65
2026-044.3%47
2026-057.3%41
2026-0616.7%18
2026-0726.7%15
2026-0823.1%13
2026-090%2

Peak month: 26.7%. Bars are scaled to that peak, not to 100%.

Which agents

AgentVendorCommitsShare
Claude CodeAnthropic83.1%
Cursor AgentAnysphere51.9%
GitHub CopilotGitHub31.2%
OpenAI CodexOpenAI20.8%

Inspect the matches

Public commit samples let you inspect the signatures behind the aggregate result.

Matched commit samples

  • 822a8fc0 on GitHub[EP] size internode TMA sender buffer for hidden up to 8192 (#1020)Claude Code
  • 3e06114f on GitHub[P2P] Add Cambricon MLU (CNRT) support for cross-node P2P RDMA (#1011)Claude Code
  • 4312141c on GitHub[Tooling] add pyproject.toml metadata and uv.lock (#434) (#1025)Claude Code
  • a8c94e48 on GitHub[CI] Skip CXI apt install when libfabric/hwloc are already present (#1041)Cursor Agent
  • 9ef7dfd4 on GitHub[EP] Add UCCL_EP_DISABLE_CPU_AFFINITY to opt out of proxy thread CPU pinniCursor Agent
  • a7176a99 on GitHub[CI] Set NCCL_SOCKET_IFNAME for GH200 EP latency test (#1034)Cursor Agent
  • 7be91f17 on GitHub[LITE]Reorganize lite to lite-ep and lite-collective (#962)GitHub Copilot
  • 8ac850bd on GitHub[UK] Unify buffer_id semantics across Transport/CCL/Python and harden OOB/GitHub Copilot
  • bcd180e7 on GitHub[EP] remove aten/torch depedency fromm uccl_ep (#784)GitHub Copilot
  • c8b236f4 on GitHub[EP] Split proxy CUDA device index from NIC local rank (#989)OpenAI Codex
  • 34bdf4f4 on GitHub[EP] Support distinct HT barrier local ranks (#985)OpenAI Codex

What entered the denominator

The result begins with commit history, removes merges and routine automation, and reports literal agent signatures as a lower bound.

Exclusion ledger

Commits in window260
Merge commits removed−0
CI / dependency bots removed−0
Considered260

Bots are excluded from the denominator entirely. Agent attribution requires a literal supported signature; no code-style inference is used.

Reproduce this observation

Run the same detector locally.

Deterministic signature matching; no code-style inference. Squash merges can remove trailers, so this result remains a measured floor.

npx qmmit-cli uccl-project/uccl

Scan another repository