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.

LLM speculative inference server for heterogeneous hardware & consumer GPUs

01Result

Luce-Org/lucebox ↗

Luce-Org/lucebox AI coding statistics: 23.7%

of 1,384 commits in February–September 15, 2026 were committed by an AI agent — a floor, not an estimate
23.7% agent · top contributor 72%22 contributors
team

A small group is using agents. Adoption has spread beyond one person but is not project-wide.

328
agent commits
1,056
other considered commits
22
agent contributors
72%
largest contributor

02Drivers

What shapes this result

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

Assisted or autonomous

327assisted
1autonomous

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
2026-0410.9%129
2026-0545%522
2026-0616.1%217
2026-079%145
2026-080%165
2026-0915%206

Peak month: 45%. Bars are scaled to that peak, not to 100%.

Which agents

AgentVendorCommitsShare
GitHub CopilotGitHub23717.1%
Claude CodeAnthropic876.3%
DevinCognition30.2%
OpenAI CodexOpenAI30.2%

03Evidence

Inspect the matches

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

Matched commit samples

  • 73b596c5 on GitHubfeat(dflash): add DeepSeek V4 Flash backend (#353)GitHub Copilot
  • d201348d on GitHubfeat(moe): add--freq and --collect-routing, train predictor of experts (#3GitHub Copilot
  • 32e0675f on GitHubfeat: hybrid routed+cold path for inline cold computeGitHub Copilot
  • 88e5e0a4 on GitHubdocs(ds4): list the qualified hybrid profile's switches in the environmentClaude Code
  • ebc43da4 on GitHubdocs(ds4): add code-prompt decode rates and the adaptive-width note to theClaude Code
  • 499ce06b on GitHubdocs(ds4): document the qualified R9700 + Strix Halo profileClaude Code
  • b3b3d2a7 on GitHubfeat(server): suspend active decode requests to bounded RAM, with recomputDevin
  • 915bddfe on GitHubrefactor(server): thread cache-type resolution through the planDevin
  • 21c1faf5 on GitHubrefactor(server): make --specla the sole SpecLA activation channelDevin
  • ebf848f3 on GitHubperf(ds4): add paged concurrent serving and batch gathered workOpenAI Codex
  • c00b8c10 on GitHubperf(ggml): optimize small HIP batches and fuse attention operationsOpenAI Codex
  • 1b11c50d on GitHubperf(laguna): align MoE router and verify paths (#455)OpenAI Codex

04Method

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 window1,801
Merge commits removed−414
CI / dependency bots removed−3
Considered1,384

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

Continue exploring

Claude Code commitsBrowse repositories with detected Claude Code signatures →GitHub Copilot commitsBrowse repositories with detected GitHub Copilot signatures →Devin commitsBrowse repositories with detected Devin signatures →OpenAI Codex commitsBrowse repositories with detected OpenAI Codex signatures →C++ AI coding statisticsCompare evidence across C++ repositories →GitHub AI commit analyzerInspect another repository with the same evidence model →
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 Luce-Org/lucebox

Scan another repository

Try