← the index
NVIDIA/Megatron-LM
17.1%
of 1,683 commits were committed by an AI agent — a floor, not an estimate
17.1% agent · top contributor 19.5%70 contributors
org-wide

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

287
agent commits
1,396
human commits
70
agent contributors
19.5%
largest contributor

Assisted or autonomous

287assisted
0autonomous

Assisted — a developer’s identity authored the commit and the agent added itself as co-author. The agent ran git commit during a working session, with a person there. Autonomous — a bot identity is the author, no human in the loop. Across the whole index, 93.7% of agent commits are assisted.

Agent share by month

Feb26
Mar26
Apr26
May26
Jun26
Jul26
Aug26
Sep26

Peak month: 30.1%. Bars are scaled to that peak, not to 100%.

Which agents

Claude CodeAnthropic28016.6%
Cursor AgentAnysphere60.4%
GitHub CopilotGitHub10.1%

Verify it yourself

Sample commits that matched. Every one links to GitHub — check the trailers.

  • a6e191fdMemory and perf fix for Tanh-Sq-Relu and FP32 logits output (#7229)Claude Code
  • 0e48cdf3fix(determinism): Separate the MoE aux-loss fusion flag so --deterministicClaude Code
  • 904ad475Add DeepSeek-V3 NVFP4 training support (#6841)Claude Code
  • 07a0de65fix(moe): free activations immediately in full recompute mode (#5197)Cursor Agent
  • f744215aAdd bump-base-image skill and update golden value comparison (#4733)Cursor Agent
  • fca1679eMXFP8 refit (#3742)Cursor Agent
  • 9ed8b0c4Fix incorrect HAVE_TE detection in multiple modules (#3763)GitHub Copilot

What was excluded

Commits in window1,791
Merge commits removed0
CI / dependency bots removed108
Considered1,683

Bots are excluded from the denominator entirely. Counting dependabot as “AI-written code” is the fastest way to produce a meaningless number.

How this was measured

Measured over February–August 2026, cloned directly. Detection is deterministic string matching against signatures agents write about themselves — no heuristics, no style analysis. Merge commits and CI bots are excluded from the denominator. The figure is a floor: squash merges strip trailers, so real usage is higher. Contributor counts are aggregate — no individual is named.

Run it yourself: npx qmmit-cli NVIDIA/Megatron-LM

Scan another repository