Standalone Commits logo

Standalone Commits

OrganizationPopular
EpicenterHQ
standalone-commits

Split work into reviewable, auditable commits ordered by dependency. Use when planning commit waves, staging changes, or deciding whether a commit is too broad, tiny, incomplete, or hard to revert.

Overview

PublisherEpicenterHQ
Repositoryepicenter
Skill namestandalone-commits
Stars
4.8K
Forks
386
Bundled files
1
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by EpicenterHQ on GitHub. Read the source before you install it.

Installation

Install the Standalone Commits AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/EpicenterHQ/epicenter/tree/main/.agents/skills/standalone-commits
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/EpicenterHQ/epicenter.git /tmp/epicenter
mkdir -p .claude/skills
cp -r /tmp/epicenter/.agents/skills/standalone-commits .claude/skills/standalone-commits
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Standalone Commits in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Standalone Commits on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Standalone Commits is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Standalone Commits

A standalone commit is a commit a reviewer can audit on its own, at that point in branch history. It may depend on earlier commits in the same branch, but it should not require future commits to explain, compile, test, or justify it.

Related Skills: See git for conventional commit messages and PR text.

Two Halves: Boundary And Order

This skill owns both halves of commit shaping. Hold them at once:

  • The review boundary (this file): is each commit worth reviewing by itself? That is the philosophy and the acceptance checks below.
  • The ordering: what order should the standalone commits happen in? For the detailed dependency-ordered wave procedure, read references/splitting-into-ordered-waves.md.

Without ordering, "standalone" drifts into polished but poorly sequenced commits. Without the boundary checks, technically ordered commits can still be hard to review.

The Standard

Each commit should answer one sentence:

This commit changes <thing> so that <outcome> because <reason>.

If the sentence needs "and also", split the commit. If the reason only becomes true in a later commit, move the work later or combine the dependent pieces.

Acceptance Checks

Before committing, verify the staged diff against these checks:

CheckQuestion
ReviewableCan a reviewer understand the purpose from this diff and message alone?
BuildableDoes the repo still typecheck, build, or pass the relevant focused test after this commit?
CompleteAre all call sites, exports, tests, docs, and fixtures needed for this concern included?
FocusedIs there exactly one behavioral, structural, or documentation concern?
AuditableCan someone inspect the before/after and see the invariant being preserved or changed?
RevertibleIf this commit were reverted, would it remove one coherent change without dragging in unrelated work?

When a commit fails one check, fix the staged set before writing the message.

What Standalone Means

Standalone does mean:

  • The commit is coherent at its point in history.
  • The message explains why the diff exists.
  • New public surface has the minimum implementation, exports, and tests needed to be credible.
  • Refactors preserve behavior and include verification when risk is non-trivial.
  • Mechanical changes are separated from semantic changes when mixing them would hide behavior.

Standalone does not mean:

  • Every commit must be useful without earlier commits in the same branch.
  • Every commit must be tiny.
  • A large generated or mechanical change must be manually split into noise.
  • The final branch cannot contain dependency chains.

Wave Planning

When ordering work into waves, plan each wave as a standalone commit:

Wave 1 claim: Add the type or contract, plus any tests that prove the new shape.
Wave 2 claim: Add the implementation that satisfies the contract.
Wave 3 claim: Move consumers onto the implementation.
Wave 4 claim: Remove old paths once no consumers depend on them.

Do not create a "setup" commit that only makes a future commit possible unless it is independently reviewable. A foundation commit is fine when it introduces a real contract, helper, migration, or test fixture with a clear reason.

Before coding a wave, write its claim. After staging it, re-read the staged diff and ask whether the claim is still honest. If not, split, combine, or move the work.

Staging Workflow

  1. Inspect the full diff.
  2. Write the one-sentence commit claim.
  3. Stage only files and hunks that prove that claim.
  4. Re-read the staged diff with git diff --staged.
  5. Run focused verification for that staged state when practical.
  6. Commit with a message that names the outcome and explains the reason when needed.

Prefer file-level staging when files cleanly map to the commit claim. Use hunk staging when one file contains multiple concerns.

Split Or Combine Heuristics

Split when:

  • A refactor and behavior change appear together.
  • A rename or move hides logic changes.
  • Tests cover a different concern than the implementation.
  • A dependency bump is mixed with feature work.
  • Reviewers would need to hold two mental models at once.

Combine when:

  • A type without its only implementation would be dead surface.
  • A helper without its call site would be speculative.
  • A test without the fix would intentionally fail in the branch.
  • A migration without the schema or reader change would break local state.

Anti-Patterns

Teaser Commit

feat(sync): add workspace sync types

Problem: The types are unused, untested, and only make sense after a later implementation commit.

Better: Include the contract-level implementation or test that proves why the types exist.

Hidden Behavior In Refactor

refactor(auth): reorganize session helpers

Problem: The diff also changes session expiry behavior. Reviewers may scan it as a move-only change.

Better: Commit the move first with no behavior change, then commit the expiry behavior change with tests.

Cleanup Hitchhiker

fix(settings): persist provider selection

Problem: The commit also renames unrelated variables and formats nearby files.

Better: Keep the fix focused. Put cleanup in a separate commit only if it is worth reviewing.

Commit Message Shape

Use the git skill for exact conventional commit formatting. For standalone commits, make the subject name the outcome, not the implementation detail.

Good:

fix(settings): persist selected provider across reloads

Weak:

fix(settings): update localStorage call

Add a body when the reason, invariant, or review boundary is not obvious from the diff.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Standalone Commits AI skill do?

Split work into reviewable, auditable commits ordered by dependency. Use when planning commit waves, staging changes, or deciding whether a commit is too broad, tiny, incomplete, or hard to revert.

Why use Standalone Commits on TypingMind?

Because you install it once and use it with any model. Standalone Commits is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Standalone Commits in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/EpicenterHQ/epicenter/tree/main/.agents/skills/standalone-commits. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Standalone Commits?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Standalone Commits?

As many as you like. As long as a model supports skills, you can use Standalone Commits with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Standalone Commits AI skill free?

It is published on GitHub by EpicenterHQ. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇