GITHUB-COPILOT PUB_DATE: 2026.01.22

STUDY: WHERE AI-AUTHORED PRS FAIL—AND HOW TO IMPROVE MERGE RATES

A large study of 33k agent-authored GitHub pull requests across five coding agents finds that documentation, CI, and build-update PRs have the highest merge suc...

Study: Where AI-authored PRs Fail—and How to Improve Merge Rates

A large study of 33k agent-authored GitHub pull requests across five coding agents finds that documentation, CI, and build-update PRs have the highest merge success, while bug-fix and performance PRs fare worst. Failed PRs typically have larger diffs, touch more files, and often fail CI; qualitative reasons include duplicate PRs, unwanted features, agent-task misalignment, and limited reviewer engagement.

[ WHY_IT_MATTERS ]
01.

Helps decide which tasks to assign to coding agents to maximize merge success.

02.

Points to concrete guardrails around PR scope, CI health, and review workflow.

[ WHAT_TO_TEST ]
  • terminal

    Pilot agents on docs/CI/build tasks with strict diff-size/file-count limits and require green CI before review.

  • terminal

    Track agent PR outcomes by task type and failure mode to tune prompts, routing, and approval rules.

[ BROWNFIELD_PERSPECTIVE ]

Legacy codebase integration strategies...

  • 01.

    Start with low-risk agent tasks and enforce PR templates, auto-labels, CODEOWNERS, and CI gating to reduce noisy PRs.

  • 02.

    Add duplicate detection and issue linking to block agent-created duplicates or misaligned features before review.

[ GREENFIELD_PERSPECTIVE ]

Fresh architecture paradigms...

  • 01.

    Adopt a small-PR policy, high test coverage, and fast CI to raise agent merge rates from day one.

  • 02.

    Define a task taxonomy (docs/CI/build vs bug-fix/perf) and route agent capabilities and approvals accordingly.