Skip to content

feat: define and implement merge conflict detection + handling workflow #4

@biz-biz-Bot

Description

@biz-biz-Bot

Problem

Open PRs can develop merge conflicts as main advances, and there is currently no automated detection or handling process. Conflicts were discovered manually on PRs #305, #329, #332, #335, #342, #344 — but nothing surfaced them proactively.

Related

The PR watcher (#350) fires on GitHub review events (comments, approvals). Merge conflicts are a different signal — they appear in a PR's mergeable / mergeability_reason fields on the GitHub API. The two concerns are orthogonal; #350 does not cover this.

Questions to resolve

  1. Who detects conflicts? Should the PR watcher also poll mergeable state? Or a separate scheduled job?
  2. Who resolves them? Smith should rebase or merge main into the branch, then push. When should this happen — on a schedule, or reactively on conflict detection?
  3. Does resolution require re-review? If the rebase changes the diff non-trivially, the assay review-loop should re-run before the PR is considered ready.
  4. What happens after merge conflicts are resolved? Labels updated, PR watcher notified?

Proposed approach (to be confirmed)

  • PR watcher (or a sibling sentinel job) polls mergeable for open PRs labeled agent-ready-for-review or agent-in-progress
  • On conflict detected: assign to Smith (label swap), Smith rebases + pushes
  • If diff changed: re-run run-review-loop.sh; proceed on APPROVED
  • Label back to agent-ready-for-review when clean

Definition of Done

A conflict on an open PR is detected automatically and resolved without manual intervention.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions