SkillAgentSearch skills...

split-and-ship

Execute an approved split plan, shipping each change group separately as its own branch and PR or as sequential commits on the current branch

Install / Use

npx skills add tobihagemann/turbo --skill split-and-ship

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

83/100

Supported Platforms

Universal

Our assessment of split-and-ship

split-and-ship scores 83/100 on our quality scale, 3264th of 4,616 Development & Engineering skills we index.

Its SKILL.md is 7.0 KB long, well organised into 13 sections and no code examples: a thorough specification that gives an agent plenty to work with.

It has 405 GitHub stars, a meaningful sign that others use it.

Substance
29/30
Structure
13/20
Description
15/15
Adoption
11/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated 12 days ago, so split-and-ship is actively maintained.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-10-05. Automated pattern scan on 2026-10-05. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

split-and-ship compared with similar skills

All 4 of these similar skills score higher than split-and-ship; compare them before choosing.

SkillScoreStarsUpdatedFormat
split-and-ship (this skill)by tobihagemann8340512d agoSKILL.md
ai-job-searchby MadsLorentzen10045.0ktodayCLAUDE.md
claude-howtoby luongnv8910041.8k5d agoCLAUDE.md
algorithmic-artby anthropics100177.9k13d agoSKILL.md
pptxby anthropics100177.9k13d agoSKILL.md

Frequently asked questions

How do I install split-and-ship?
Run npx skills add tobihagemann/turbo --skill split-and-ship. The install tabs above show the steps for each supported agent.
Which AI agents does split-and-ship work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is split-and-ship safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is split-and-ship still maintained?
The repository was last updated 12 days ago, so split-and-ship is actively maintained.

name: split-and-ship description: "Execute an approved split plan, shipping each change group separately as its own branch and PR or as sequential commits on the current branch. Use when the user asks to "split and ship", "ship the split plan", "create separate PRs", or "split changes into branches"."

Split and Ship

Ship an approved split plan so each change group becomes its own reviewable unit, either as a separate branch and PR or as a sequential commit on the current branch.

Context

A split plan must exist in the conversation. The plan specifies an ordered list of groups, each with a name, file list, and any dependencies on earlier groups.

Task Tracking

At the start, use TaskCreate to create a task for each phase:

  1. Choose realization path
  2. Ship the groups
  3. Summarize

Step 1: Choose Realization Path

Detect the repository state:

  • Default branch: gh repo view --json defaultBranchRef --jq '.defaultBranchRef.name'
  • Current branch name, and whether it tracks an upstream
  • Whether a PR already exists for the current branch (gh pr view)

Sample the prevailing workflow from recent default-branch history (git log --first-parent origin/<default-branch> -n 30 --pretty=%s): judge whether changes mostly land through pull requests (merge-PR commits or (#N)-suffixed squash commits) or are committed directly to the default branch.

Output a one-line summary of the detected state as text. Then use AskUserQuestion to choose how to ship the groups:

  • Separate branches and PRs — one branch and PR per group (Step 2)
  • Commit each group and push — one commit per group on the current branch, then push (Step 3)
  • Commit each group only — one commit per group on the current branch, no push (Step 3)

Recommend the option that fits this repo by listing it first and labeling it (Recommended): a PR-based history recommends separate branches and PRs; a direct-commit history recommends committing each group on the current branch.

If the user declines (chooses the free-form "Other" option or asks to abort), leave the changes staged and do not commit.

Step 2: Ship as Separate Branches and PRs

Take this step only when the user chose separate branches and PRs.

Prepare the working tree

Stash all changes including untracked files (git stash --include-untracked) without unstaging first.

Verify git stash list shows the saved changes and git status --short lists nothing the stash should have taken before proceeding. Entries for a submodule, a nested repository, or files a stashed ignore rule had hidden may remain. When either check fails, stop and report where the changes now are.

Ship each group

Use TaskCreate to create a task for each group. Process groups in order.

For each group:

  1. Determine branch: If the current branch already has a PR and this group's changes align with the PR's purpose, stay on the current branch. Otherwise, use AskUserQuestion to confirm the proposed branch name and create it from the appropriate base:
    • Group that builds on an earlier group: branch from that group's branch (stacked)
    • Group with no dependencies: branch from the default branch (independent)
  2. Restore and stage this group's files from the stash's staged snapshot (git checkout 'stash^2' -- <files> restores and stages in one operation). For files with hunks belonging to different groups, restore the file, use Edit to remove the unwanted hunks, then re-stage it (git add <file>): for an independent group, remove every other group's hunks; for a stacked group, remove only later groups' hunks (earlier groups' hunks are already in its base). After committing, reset the working tree (git checkout -- .) to clean up before the next group.
  3. Commit and push: run the /commit-rules skill to load commit message rules, commit the staged changes following them, then git push
  4. Create or update PR:
    • Staying on existing branch with a PR: run the /update-pr skill
    • New branch: run the /create-pr skill targeting the appropriate base (default branch for independent groups, previous group's branch for stacked groups)

Clean up and summarize

  1. Drop the stash
  2. Check out the last created branch
  3. Output a summary table: group name, branch, PR URL, and base branch. For a group whose PR was not posted, print the handed-over body file path in place of the URL.

Then use the TaskList tool and proceed to any remaining task.

Step 3: Ship as Commits on the Current Branch

Take this step only when the user chose to commit each group. This path stays on the current branch and creates no branches or PRs.

Prepare the working tree

Stash all changes including untracked files (git stash --include-untracked) without unstaging first.

Verify git stash list shows the saved changes and git status --short lists nothing the stash should have taken before proceeding. Entries for a submodule, a nested repository, or files a stashed ignore rule had hidden may remain. When either check fails, stop and report where the changes now are.

Commit each group

Run the /commit-rules skill to load commit message rules. Use TaskCreate to create a task for each group. Process groups in order.

For each group:

  1. Restore and stage this group's files from the stash's staged snapshot (git checkout 'stash^2' -- <files> restores and stages in one operation). For files with hunks belonging to different groups, restore the file, use Edit to remove the hunks that belong to later groups, then re-stage it (git add <file>).
  2. Commit the staged changes with a message following the loaded rules. If a commit hook modifies files, re-stage them before retrying.
  3. Reset the working tree (git checkout -- .) to clean up before the next group.

Push and summarize

  1. If the chosen option includes pushing, push the current branch's remote (git push)
  2. Drop the stash
  3. Output a summary table: group name and commit subject

Then use the TaskList tool and proceed to any remaining task.

Rules

  • Run the /commit-rules skill before every commit; do not commit without loading it first.
  • Never lose uncommitted work. Both paths stash all changes before shipping. Before dropping the stash, restore and report every stashed change that no group shipped, rather than discarding it: re-apply the unstaged changes with git diff 'stash^2' stash | git apply --3way --allow-empty, restore untracked files with git checkout 'stash^3' -- <files>, then unstage everything with git reset -q. Report any hunk that does not apply. If any step fails (commit hook, push, PR creation), stop and report the failure, which groups have been shipped, and that the stash still contains all changes for recovery. A group whose PR was handed over for editing rather than posted is not a failure; record it in the summary and continue.
  • For stacked groups, the PR description should note the dependency chain.
  • Don't reference .turbo/ content (filenames, acceptance criteria, step numbers, headings) in branch names. .turbo/ is gitignored, so these references would be opaque to anyone reading without local copies.

Related Skills

View on GitHub
GitHub Stars405
CategoryDevelopment
Updated12d ago
Forks31

Languages

Python

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions