• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Agent-generated pull requests reportedly grew ninefold in size over eight months

    A post citing GitHub data links the growth to a rethink of how code gets reviewed and deployed, with software monitoring possibly affected too.

    Leo BoytsovLB
    Gergely OroszGO
    Kyle DaigleKD
    5 Sources, ,

    TLDR

    A September 16, 2026 post citing GitHub data says agent-generated pull requests—proposed code changes—grew ninefold in size over the preceding eight months. The author says the growth shows no signs of slowing and argues that it is prompting a rethink of code reviews, deployments and possibly observability: how teams monitor software behavior.

    Combined views

    71.2K

    5 Sources, first seen 21d ago

    Combined views

    71.2K

    5 Sources, first seen 21d ago

    430 likes
    21d ago
    first seen 21d ago
    430 likes
    57 comments
    154 saves
    44 reposts
    57 comments
    154 saves
    44 reposts

    Sentiment

    Positive4.9%95.1%Negative

    Summary

    Positive accounts endorse Copilot for scaling automated reviews, while negative accounts argue 9x-larger agent PRs make traditional diff review unsustainable and demand smaller boundaries with executable criteria.

    Based on 45 sentiment-bearing replies from 41 accounts across 2 conversations.

    Sentiment

    Positive4.9%95.1%Negative

    Summary

    Positive accounts endorse Copilot for scaling automated reviews, while negative accounts argue 9x-larger agent PRs make traditional diff review unsustainable and demand smaller boundaries with executable criteria.

    Based on 45 sentiment-bearing replies from 41 accounts across 2 conversations.

    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    5 Sources

    Gergely Orosz@GergelyOroszFresh data from GitHub: Agent-generated PRs have exploded in size. 9x the last 8 months (!!) No signs of slowing down. This is why everyone is rethinking code reviews, deploys, possibly even o11y thanks @kdaigle + team for the stat! (Slide from my talk yesterday)21d
    Kyle Daigle@kdaigleThe best projects are using more automated reviews, validation/smoke testing, auto-rollbacks than ever before to keep up with the pace. PRs are growing incredibly quickly, we see Copilot Code Review being used a TON more even if you use different agents to create the PR.21d
    GitHub Enterprise@GitHubEntAI coding tools increase developer output. But growing pull request backlogs can easily overwhelm engineering teams, contributing to what’s known as “review fatigue.” Here’s how to calibrate your workflows for agentic development. Improve prompts, use stacked pull requests, and recognize how the developer role is changing. https://github.com/resources/insights/agentic-engineering-system20d
    Leo Boytsov@srchvrsThere is just one tiny catch: Nobody (and nothing) can really verify that these gigantic PRs work. Your testing pipeline is dominated by vibe-coded tests, frequently written by the same model that generated the implementation (to pass the code). Serious integration testing? Production-like testing? End-to-end verification? Maybe eventually. ↩️19d

    5 Sources

    Gergely Orosz@GergelyOroszFresh data from GitHub: Agent-generated PRs have exploded in size. 9x the last 8 months (!!) No signs of slowing down. This is why everyone is rethinking code reviews, deploys, possibly even o11y thanks @kdaigle + team for the stat! (Slide from my talk yesterday)21d
    Kyle Daigle@kdaigleThe best projects are using more automated reviews, validation/smoke testing, auto-rollbacks than ever before to keep up with the pace. PRs are growing incredibly quickly, we see Copilot Code Review being used a TON more even if you use different agents to create the PR.21d
    GitHub Enterprise@GitHubEntAI coding tools increase developer output. But growing pull request backlogs can easily overwhelm engineering teams, contributing to what’s known as “review fatigue.” Here’s how to calibrate your workflows for agentic development. Improve prompts, use stacked pull requests, and recognize how the developer role is changing. https://github.com/resources/insights/agentic-engineering-system20d
    Leo Boytsov@srchvrsThere is just one tiny catch: Nobody (and nothing) can really verify that these gigantic PRs work. Your testing pipeline is dominated by vibe-coded tests, frequently written by the same model that generated the implementation (to pass the code). Serious integration testing? Production-like testing? End-to-end verification? Maybe eventually. ↩️19d