← the index
BerriAI/litellm
21.6%
of 8,796 commits were committed by an AI agent — a floor, not an estimate
21.6% agent · top contributor 21.1%122 contributors
org-wide

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

1,898
agent commits
6,898
human commits
122
agent contributors
21.1%
largest contributor

Assisted or autonomous

1727assisted
171autonomous

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

Peak month: 34.8%. Bars are scaled to that peak, not to 100%.

Which agents

Claude CodeAnthropic117613.4%
Cursor AgentAnysphere4605.2%
DevinCognition3083.5%
GitHub CopilotGitHub260.3%
OpenAI CodexOpenAI70.1%
Other / generic agentVarious50.1%

Verify it yourself

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

  • f9704497feat(helm): add startupProbe and hpa.behavior to the componentized chart (Claude Code
  • add095b4Fix: ComplexityRouter should not score system prompt text for code/technicClaude Code
  • 1911269dfix(router): never price a strategy-router alias (#36691)Claude Code
  • d30e933eci: wire credential_endpoints tests into the proxy endpoints job (#36187)Cursor Agent
  • b792fd7ctest(router): cover PatternMatchRouter.remove_deployment for router code cCursor Agent
  • 836bf080fix(router): keep team wildcard routers fresh and prioritize them over gloCursor Agent
  • 1abde192docs(claude): require ReadOnly on every TypedDict field (LIT012)Devin
  • 6e7984e5fix(proxy): requeue spend logs when the DB write fails with a transport erDevin
  • e46f2ca6fix(dashscope): honor the model reasoning rate when a tier omits output raDevin
  • 5677bc23fix(proxy/batches): resolve managed unified input_file_id to storage_url wGitHub Copilot
  • 0a17c7c3feat: add LiteLLM Rust workspace with Mistral OCR bridge (#31033)GitHub Copilot
  • 1ccc1e5bchore: litellm oss staging160626 (#30527)GitHub Copilot
  • b4a4277afix(ui): align spend and budget columns (#35176)OpenAI Codex
  • cb888364Add incomplete response error propagation testOpenAI Codex
  • bd5c39c4Log incomplete details in background streamingOpenAI Codex
  • b84f7f82Litellm oss staging (#29492)Other / generic agent
  • e8fcb012Litellm OSS Staging (#29161)Other / generic agent
  • 58cd6f68test: fix misleading precedence test per review feedbackOther / generic agent

What was excluded

Commits in window11,687
Merge commits removed2,874
CI / dependency bots removed17
Considered8,796

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 BerriAI/litellm

Scan another repository