EmbodiChain Pull Request Creator

This skill guides you through creating a pull request that follows the EmbodiChain project's PR template and contribution guidelines.

Usage

Invoke this skill when:

• You have completed a feature, bug fix, or other change and want to create a PR
• You want to split a large, dependency-ordered change into stacked PRs
• You want to ensure the PR follows the project's conventions
• You need help drafting a proper PR description

Steps

1. Check Current State

First, check the current git status and changes:

git status
git diff HEAD

2. Determine Change Type

Based on the changes made, select one of these PR types:

• Bug fix - Non-breaking change which fixes an issue
• Enhancement - Non-breaking change which improves an existing functionality
• New feature - Non-breaking change which adds functionality
• Breaking change - Existing functionality will not work without user modification
• Documentation update

3. Draft the PR Description

Write a description that includes:

• Summary: A clear, concise summary of the change
• Issue reference: Which issue is fixed (e.g., "Fixes #123")
• Motivation and context: Why this change is needed
• Dependencies: List any dependencies required for this change

4. Review Test Necessity and Select Proportional Validation

Before validation and publication, apply the shared test necessity review. Inspect the full proposed PR delta from the merge base with its intended target branch, plus staged, unstaged, and untracked changes to be included. git diff HEAD alone misses committed additions. Include new or extended cases and supporting fixtures in existing files, not only newly added test files.

Within the task's authorized scope, remove or merge confirmed redundant additions and simplify unjustified implementation coupling. Limit cleanup to this change and preserve explicitly requested scenarios. Run the affected tests after any edits. Briefly state the coverage basis and any meaningful cleanup in the PR's validation summary, including existing coverage when no new tests are needed; do not create a per-test justification ledger.

Inspect the changed files and validate only the affected behavior. Do not run the full test suite automatically.

• Workflow-only changes: run actionlint on the changed workflows and any directly related workflow-script tests.
• Documentation-only changes: run the relevant docs checks or build; do not run Python tests unless executable examples or docs tooling changed.
• Isolated Python changes: run the nearest matching test module or package.
• Shared infrastructure or cross-package changes: run affected package tests and focused integration tests.
• Run the full suite only for broad cross-cutting changes, global dependency or test-configuration changes, release-critical changes that cannot be validated narrowly, or when the user explicitly requests it.

If a validation command is likely to take more than two minutes, state why it is needed before starting it. Honor an explicit user request to skip or narrow tests, and record skipped checks honestly in the PR description.

Use the pre-commit-check skill for the detailed selection and reporting rules.

5. Run Code Formatting

Before creating the PR, ensure code is formatted:

black .

If formatting changes were made, commit them first:

git add -A
git commit -m "Format code with black"

Then run the read-only public API documentation gate used by CI:

python docs/scripts/check_api_docs.py

If it reports missing public exports, invoke $update-api-docs to generate or update the relevant Sphinx entries and descriptions, then rerun the checker. Keep generation in that specialized skill; do not add placeholder API docs in the PR workflow merely to make the gate pass.

6. Create or Update Branch

For a single PR, create a feature branch if needed:

git checkout -b <branch-name>

Recommended branch naming:

• fix/<description> - for bug fixes
• feat/<description> - for new features
• enhance/<description> - for enhancements
• docs/<description> - for documentation changes

For a large change that can be split into small, dependency-ordered layers, follow the Stacked Pull Request Workflow instead of steps 6 through 9. Keep unrelated or independent changes in separate PRs rather than forcing them into a stack.

7. Commit Changes

Commit with a clear message following conventional commits format:

git commit -m "type(scope): brief description

Detailed description of the change."

8. Push to Remote

git push -u origin <branch-name>

9. Create the PR

Use the gh CLI with the proper PR template:

gh pr create --title "<PR Title>" --body "<PR Body>"

Stacked Pull Request Workflow

Use a stack when later changes depend on earlier changes but each layer can be reviewed as a focused PR. Put shared types, schemas, and other foundations in lower layers; put their dependents in higher layers. Keep every branch and PR in the same repository because GitHub does not support cross-fork stacks. If the contribution must come from a fork, use a single PR or separate non-stacked PRs.

[!NOTE] GitHub stacked pull requests are in public preview and may change. They are available on GitHub, GitHub Mobile, and through the gh stack extension, but not in GitHub Desktop.

1. Install and Authenticate the CLI

Use GitHub CLI 2.90.0 or later and Git 2.20 or later. Authenticate, then install the official extension if it is not already installed:

gh auth login
gh extension install github/gh-stack

2. Design the Stack

Define the trunk and order the layers from foundational to dependent. The bottom PR targets the trunk, usually main, and each PR above it targets the branch immediately below it:

feat/ui          -> PR #3 (base: feat/api)   <- top
feat/api         -> PR #2 (base: feat/core)
feat/core        -> PR #1 (base: main)       <- bottom
main             <- trunk

Ensure each PR contains only the diff for its layer. If code in one layer depends on another, place the dependency in the same layer or a lower layer.

3. Create and Commit the Layers

Initialize the bottom branch, commit its focused change, then add branches from the bottom upward:

gh stack init --base main feat/core
# Make, stage, validate, and commit the foundational change.

gh stack add feat/api
# Make, stage, validate, and commit the dependent API change.

gh stack add feat/ui
# Make, stage, validate, and commit the dependent UI change.

gh stack view

Apply the test necessity review and proportional validation in step 4 to every affected layer, using that layer's own target branch as its base. Keep the commits and working tree for each layer complete before adding the next branch.

4. Publish or Update the Stack

Use the interactive submit flow so each PR can receive an EmbodiChain title and body. This command pushes the branches, creates or updates their PRs, and links them as a stack:

gh stack submit

Use gh stack push only when the branches should be pushed without creating or updating PRs. Use gh stack submit --auto only when generated titles and draft PRs are acceptable; add --open to make them ready for review.

After the remote stack or trunk changes, synchronize the entire chain:

gh stack sync

After amending a lower layer, cascade the new commit history through its dependents, then update the remote PRs:

gh stack rebase --no-trunk
gh stack submit

If a rebase stops on conflicts, resolve and stage them, then run gh stack rebase --continue; use gh stack rebase --abort to restore the pre-rebase state. Preserve the dependency order and rerun validation for every affected layer before submitting again.

5. Describe Every Layer

Apply the standard PR template to every PR and add this stack context near the top of each description:

## Stack

- Layer: <position>/<total>
- Base: `<base-branch>`
- Depends on: <lower PR URL or "none">
- Followed by: <upper PR URL or "none">

Explain what can be reviewed in that layer and report its validation separately. Use Fixes #<issue> only on the layer whose independent merge resolves the issue; use Refs #<issue> on prerequisite layers. Apply appropriate repository labels to every PR in the stack.

6. Review and Merge

Review every layer independently. GitHub applies the trunk's merge requirements, branch protections, CODEOWNER approvals, and applicable PR-triggered CI checks to every layer, including PRs that do not directly target the trunk.

Merge from the bottom upward:

• Merge the top PR to land the entire stack.
• Merge a middle PR to land it and every lower layer; higher PRs remain open and GitHub automatically rebases and retargets them.
• Merge the bottom PR to land only the first layer.

Use gh stack merge to select and merge the desired portion. The operation is all-or-nothing for the selected layers and cannot bypass merge requirements:

gh stack merge

10. Select and Apply Labels

After creating the PR, select proper labels from the repository label list and apply them.

First, list available labels:

gh label list --limit 100

Then choose labels based on change type and scope. Typical mapping:

• Bug fix: bug
• Improvement to an existing capability: enhancement
• New feature: feature
• Documentation update: docs
• Structural change: refactor; performance work: performance
• Add breaking for user-visible incompatible changes, independently of change type.
• Select the affected areas from the current repository labels. Preserve the exact label motion gen for motion planning, generation, retiming and expansion.

Keep these ownership boundaries explicit:

• task: concrete tasks, deployments and objectives; task-program: language, semantics, compiler, runtime and integrations; gym: environment framework.
• atomic-skill: Atomic Skill contracts and execution; kinematics: FK/IK, Jacobians and reachability; physics: dynamics and physics solvers.
• gen-sim: scene generation/editing and SimReady; scene: shared scene composition/layout/state services; assets: resource files and catalogs; sim-object: runtime simulation entities and their lifecycle.
• data-pipeline: collection, ODS, sampling, recording, persistence, conversion and analysis. The former data and dataset labels are consolidated here.
• model: model architectures, inference and weights; sl and rl: the corresponding training systems; compute: shared numerical algorithms.
• agent: runtime agents, MLLM and MCP integration; devtools: repository development skills, agent context and contribution tooling.
• rendering: image generation and renderer behavior; visualization: viewers, markers, previews and interactive debugging. Use these instead of interaction.
• benchmark: evaluation protocols, suites, metrics and reports; performance: optimization work. Use both when both scopes change.
• dexsim: SDK interfaces and version integration, rather than every simulation change; robot, sensor and toolkit: their corresponding runtime/tool owners.

Use ci for build/release automation and dependencies for dependency changes. Apply skip-release-notes only when explicitly excluding the change from release notes. Draft state and dependency links describe PR progress; area labels do not claim that a planned capability has shipped.

Apply labels to the PR:

gh pr edit <pr-number> --add-label "bug" --add-label "gym"

If needed, remove incorrect labels:

gh pr edit <pr-number> --remove-label "<label-name>"

PR Template

Use this template for the PR body:

## Description

<!-- Clear summary of the change -->

This PR [briefly describe what the PR does].

<!-- Include motivation and context if needed -->
[Add any relevant motivation and context here].

<!-- List dependencies if applicable -->
Dependencies: [list any dependencies required]

<!-- Reference the issue -->
Fixes #<issue-number>

## Type of change

<!-- Select one and delete the others -->
- [x] Bug fix (non-breaking change which fixes an issue)
- [ ] Enhancement (non-breaking change which improves an existing functionality)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (existing functionality will not work without user modification)
- [ ] Documentation update

## Screenshots

<!-- Attach before/after screenshots if applicable -->

## Checklist

- [x] I have run the `black .` command to format the code base.
- [ ] I reviewed affected documentation and agent context, updated it where needed, or explained why no update was needed.
- [ ] Public API changes are reflected in the API docs (`python docs/scripts/check_api_docs.py`), if applicable
- [ ] I verified appropriate behavioral coverage, reviewed the necessity of added or extended tests, and explained when no new tests were needed.
- [ ] Dependencies have been updated, if applicable.

PR Title Guidelines

• Keep titles short (under 70 characters)
• Use present tense and imperative mood
• Examples:
  • "Fix KeyError when 'add' mode not present in observation_manager"
  • "Add support for XYZ sensor"
  • "Improve contact sensor data buffer"

Quick Reference

Notes

• Keep PRs small and focused. Large PRs are harder to review and merge.
• It's recommended to open an issue and discuss the design before opening a large PR.
• The checklist in the PR template should be completed honestly.
• For current stacked PR behavior, consult the public preview announcement, GitHub stacked PR overview, and CLI command reference.