From blueprint-dev
Reference for how the built-in /batch command integrates with blueprint-dev workflows — parallel codebase-wide changes using worktrees with project context.
How this skill is triggered — by the user, by Claude, or both
Slash command
/blueprint-dev:batch-integrationThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
How the built-in `/batch` command fits into blueprint-dev workflows.
How the built-in /batch command fits into blueprint-dev workflows.
/batch is a built-in Claude Code command that performs codebase-wide changes in parallel:
/simplify, and creates a PRThis is ideal for repetitive, pattern-based changes that touch many files but follow the same transformation logic.
| Signal | Use /bp:build | Use /bp:batch |
|---|---|---|
| Files affected | 1-15 | 10+ (same pattern) |
| Change pattern | Unique per file | Same transformation repeated |
| Dependencies between changes | Yes | No (independent units) |
| Examples | Add a new feature, refactor a module | Rename API field across all endpoints, migrate from v1 to v2 SDK, add logging to all controllers |
| Planning needed | Yes (requirements, architecture) | Minimal (pattern is known) |
| Review strategy | Single /bp:review | /bp:review per PR or batch review |
/bp:batch is a blueprint-dev wrapper around the built-in /batch. It adds project context:
| Feature | Raw /batch | /bp:batch |
|---|---|---|
| Stack conventions | No | Reads .blueprint/stack-profile.json |
| Project rules | No | Reads CLAUDE.md |
| Plan/ADR context | No | Reads .blueprint/plans/, .blueprint/adrs/ |
| Batch manifest | No | Creates .blueprint/batches/{date}-{slug}.md |
| Post-batch review | No | Offers /bp:review on completed PRs |
| /simplify | Each worker runs it | Same (inherited from /batch) |
/batch naturally aligns with trunk-based development:
This is better TBD practice than a single massive PR touching 50 files.
/bp:batch "migrate all API endpoints from v1 to v2 SDK"
│
├── Step 1: Read project context (stack profile, CLAUDE.md, plans)
├── Step 2: Invoke built-in /batch with project context
│ ├── /batch decomposes into units
│ ├── Spawns parallel worktree workers
│ ├── Each worker: implement → /simplify → create PR
│ └── Returns summary
├── Step 3: Create batch manifest at .blueprint/batches/
└── Step 4: Show summary, offer /bp:review on PRs
Each batch run creates a manifest at .blueprint/batches/{date}-{slug}.md:
# Batch: {description}
**Date**: {YYYY-MM-DD}
**Units**: {count}
**PRs created**: {count}
## Units
| # | Description | PR | Status |
|---|-------------|-----|--------|
| 1 | {unit desc} | #{pr} | created |
| 2 | {unit desc} | #{pr} | created |
## Context Used
- Stack profile: .blueprint/stack-profile.json
- CLAUDE.md conventions applied
/bp:batch instead of /bp:build/bp:batch as an alternativenpx claudepluginhub dlabs/claude-marketplace --plugin blueprint-devDecomposes large-scale changes into independent units and spawns parallel agents in isolated worktrees. Use for migrations, refactors, codemods, and changes touching 10+ files.
Decomposes large tasks (migrations, multi-issue fixes, big features) into parallel work packages with quality gates. Useful for reducing PR cycle time and avoiding merge conflicts.
Orchestrates parallel git worktree-isolated background agents to implement multiple pending features simultaneously. Handles validation, spawning, tracking, and handoff to merge. Requires prior planning.