stack-workflows（堆栈工作流程）

GitButler Stack Workflows

Dependent branches → anchor-based stacking → reviewable chunks.

<when_to_use>

• Sequential dependencies (e.g., refactor → API → frontend)
• Large features broken into reviewable chunks
• Granular code review (approve/merge early phases independently)
• Post-hoc stack organization after exploratory coding

NOT for: independent parallel features (use virtual branches), projects using Graphite stacking

</when_to_use>

<stack_vs_virtual>

Stacked vs Virtual Branches

Stacked branches = virtual branches split into dependent sequence. Default: Virtual branches are stacks of one.

</stack_vs_virtual>

Creating Stacks

# Base branch (no anchor)
but branch new base-feature

# Stacked branch (--anchor specifies parent)
but branch new child-feature --anchor base-feature

# Third level
but branch new grandchild-feature --anchor child-feature

Result: base-feature ← child-feature ← grandchild-feature

Short form: -a instead of --anchor

but branch new child -a parent

Stack Patterns

Feature Dependency Stack

# Auth foundation
but branch new auth-core
but commit auth-core -m "feat: add authentication core"

# OAuth layer depends on auth core
but branch new auth-oauth --anchor auth-core
but commit auth-oauth -m "feat: add OAuth integration"

# Social login depends on OAuth
but branch new auth-social --anchor auth-oauth
but commit auth-social -m "feat: add social login"

Refactoring Stack

# Extract utilities
but branch new refactor-extract-utils
but commit refactor-extract-utils -m "refactor: extract common utilities"

# Update consumers
but branch new refactor-use-utils --anchor refactor-extract-utils
but commit refactor-use-utils -m "refactor: use extracted utilities"

# Clean up
but branch new refactor-cleanup --anchor refactor-use-utils
but commit refactor-cleanup -m "refactor: remove deprecated code"

Deep Stack (5 levels)

but branch new db-schema
but branch new data-access --anchor db-schema
but branch new business-logic --anchor data-access
but branch new api-endpoints --anchor business-logic
but branch new frontend-integration --anchor api-endpoints

<post_hoc>

Post-Hoc Stack Organization

Problem: Created branches independently, now want to stack them.

Solution: Recreate with correct anchors:

# Current: three independent branches
# feature-a, feature-b, feature-c

# Stack feature-b on feature-a
but branch new feature-b-stacked --anchor feature-a
commit_sha=$(but log | grep "feature-b:" | head -1 | awk '{print $1}')
but rub $commit_sha feature-b-stacked
but branch delete feature-b --force

# Stack feature-c on feature-b-stacked
but branch new feature-c-stacked --anchor feature-b-stacked
commit_sha=$(but log | grep "feature-c:" | head -1 | awk '{print $1}')
but rub $commit_sha feature-c-stacked
but branch delete feature-c --force

</post_hoc>

<pr_workflow>

PR Preparation for Stacks

GitButler CLI lacks native PR submission. Use GitHub CLI:

# Push branches
git push -u origin base-feature
git push -u origin dependent-feature

# Create PRs with correct base branches
gh pr create --base main --head base-feature \
  --title "feat: base feature" \
  --body "First in stack"

gh pr create --base base-feature --head dependent-feature \
  --title "feat: dependent feature" \
  --body "Depends on base-feature PR"

GitHub Settings:

• Enable automatic branch deletion after merge
• Use Merge strategy (recommended) — no force pushes needed
• Merge bottom-to-top (sequential order)

</pr_workflow>

Stack Reorganization

Squashing Within Stack

newer_commit=$(but log | grep "newer" | awk '{print $1}')
older_commit=$(but log | grep "older" | awk '{print $1}')
but rub $newer_commit $older_commit

Moving Commits Between Stack Levels

commit_sha=$(but log | grep "specific commit" | awk '{print $1}')
but rub $commit_sha correct-branch

Splitting a Branch

# Original has multiple features
but branch new second-feature --anchor original-branch
commit_sha=$(but log | grep "second feature commit" | awk '{print $1}')
but rub $commit_sha second-feature

Stack Navigation

Note: Virtual branches don't need checkout — all branches active simultaneously.

# View full stack structure
but log

# Work on any branch directly (no checkout needed)
but commit base-feature -m "update base"
but commit dependent-feature -m "update dependent"

# JSON for programmatic analysis
but --json log | jq '.[] | .branchDetails[] | {name, baseCommit}'

ALWAYS:

• Create stacks with --anchor from the start
• Merge stacks bottom-to-top (base first, dependents after)
• Snapshot before reorganizing: but snapshot --message "Before stack reorganization"
• Keep each level small (100-250 LOC) for reviewability
• Delete empty branches after reorganization

NEVER:

• Skip stack levels when merging
• Stack independent, unrelated features (use virtual branches)
• Create deep stacks (5+ levels) without good reason
• Forget anchor when creating dependent branches

Common Issues

Recovery

# Recreate branch with correct anchor
but branch new child-stacked --anchor parent
commit_sha=$(but log | grep "child:" | head -1 | awk '{print $1}')
but rub $commit_sha child-stacked
but branch delete child --force

<best_practices>

Best Practices

Planning

• Start simple: 2-3 levels max initially
• Single responsibility per level
• Only stack when there's a real dependency

Maintenance

• Run but log regularly to verify structure
• Commit to correct branches immediately
• Clean up empty branches

Communication

• Clear commit messages explaining why stack level exists
• Descriptive names indicating stack relationship
• Share but status when coordinating

</best_practices>

• version-control skill — core GitButler workflows
• complete-branch skill — merging to main
• multi-agent skill — multi-agent coordination
• GitButler Stacks Docs