agentleFS
cognitedatacognitedata/builder-skills/skills/pull-changes-resolve-conflicts/SKILL.md

pull-changes-resolve-conflicts

Skill6 starsChanged 8 days agoApache-2.0View on GitHub

Standard workflow for pulling updates from main or other branches on multi-contributor projects (including Flows apps) without silently discarding work. Guides fetching/merging, requires listing merge conflicts explicitly, analyzing ours vs theirs using conversation history and repo context, presenting prioritized recommendations, and obtaining user answers before editing conflict markers or completing the merge. Triggers: pull main, merge main, merge origin, rebase, merge conflict, unmerged paths, both modified, integrate branch, sync with main, git merge abort, resolve conflicts, UU status, theirs vs ours, feat branch update.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Prefer git fetch then git merge origin/main (or the named branch) unless the user asked for rebase.

    line 40
  2. Distinctive:

    If merge is already in progress, run git status and list every unmerged file.

    line 41
  3. Distinctive:

    Prefer small, surgical edits.

    line 76

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Pull changes & resolve merge conflicts

Use this skill whenever integrating another branch (usually main) into the current feature branch, or when git status shows unmerged paths after a merge or rebase. Applies to any Git-based team workflow; Flows/React apps are a common case where conflicts cluster in app shells and shared libraries.

Goals

  • Preserve intentional work on the current branch; do not assume “main wins” or “ours wins” without analysis.
  • Make trade-offs visible to the user before any conflict resolution edits.
  • Order discussion by impact: structural / feature / API / data-model changes before styling, copy, or spacing.

Hard rules

  1. No silent resolution — Do not remove <<<<<<< / ======= / >>>>>>> or run git add on conflicted files until the user has agreed to the plan (or explicitly says “use your recommendations”).
  2. Stop at conflicts — If a merge or rebase introduces conflicts, pause and report; do not bulldoze through large files by picking one side wholesale unless the user explicitly requests that.
  3. Prioritize impact — When presenting conflicts, group and order roughly as:
    • P0 — Structural / product: removed routes, deleted modules, dropped features, changed data or API contracts, SDK or schema changes, auth or routing shell.
    • P1 — Behavior: logic, hooks, queries, filters, error handling, loading states.
    • P2 — UI structure: layout regions, new or removed sections, navigation.
    • P3 — Presentation: tokens, spacing, class names, copy tweaks.

Workflow

1. Fetch and integrate (or diagnose)

  • Prefer git fetch then git merge origin/main (or the named branch) unless the user asked for rebase.
  • If merge is already in progress, run git status and list every unmerged file.

2. Report conflicts to the user (explicit)

Output a clear list:

  • Branch state: current branch, target branch (e.g. origin/main), merge vs rebase.
  • Unmerged files: paths only, then optionally git diff --name-only --diff-filter=U.
  • Per file (short): one line on what diverged (e.g. “AlertsPage — layout + new data scope”) if inferable from paths and git diff without resolving.

3. Analyze before editing

Use all of:

  • Conversation history — What was the user or team trying to ship on this branch?
  • Repo signals — Product or architecture docs if present (e.g. PRD.md), recent commits on the current branch, file ownership (e.g. large feature module vs shared lib/).
  • Conflict hunks — git show :2:path (ours) vs git show :3:path (theirs) during merge, or read conflict markers; identify duplicated vs orthogonal changes.

Classify each conflicted area as:

  • Orthogonal — safe to combine (e.g. import sort + new prop).
  • Overlapping — must choose or manually merge (same lines).
  • Corruption risk — duplicated blocks (common after bad merges); flag and recommend reconstructing from one side then re-applying the other side’s intent manually.

4. Recommendations + questions (required)

Present to the user:

  1. Summary table or bullets — file → recommended side or “manual merge” → one-line why.
  2. Ordered by P0 → P3 — call out anything that removes a feature or changes public behavior first.
  3. Explicit questions — anything ambiguous (e.g. “Keep main’s global behavior or the branch’s scoped variant?”).
  4. Ask for direction — e.g. “Reply with: (a) follow recommendations, (b) keep branch for file X, (c) keep main for file Y, (d) abort merge.”

Only after the user confirms (or gives a precise mapping), apply resolutions:

  • Prefer small, surgical edits; preserve both sides’ intent when possible.
  • Re-run git status; ensure no conflict markers remain; run tests or lint the user cares about for touched areas.

5. If the user wants to abort

  • git merge --abort or git rebase --abort as appropriate; confirm they lose in-progress integration state for that operation.

Anti-patterns (do not do)

  • Picking --ours / --theirs on the whole repo without user approval.
  • “Resolving” by deleting a feature branch’s work because main touched the same file.
  • Hiding conflict lists inside a long code dump without a short executive summary.
  • Fixing low-impact style conflicts first while leaving P0 decisions implicit.

Quick reference

  • Ours vs theirs (merge): stage 2 = current branch (HEAD), stage 3 = incoming (MERGE_HEAD). Verify with git checkout --conflict=merge <file> if needed.
  • Typical high-touch paths in full-stack or Flows apps: root app shell, top navigation, route modules, and shared lib/ or hooks/.

For optional command snippets and a merge message template, see reference.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.