By arittr
Orchestrates spec-anchored development workflows by decomposing feature specs into execution plans, then running tasks in parallel using git worktrees and git-spice branch stacking with automatic sequential/parallel phase analysis and recovery guides. Includes TDD workflow enforcement and environment validation.
Spec-anchored development workflow with automatic sequential/parallel orchestration using superpowers skills and git-spice.
Execute implementation plan with automatic sequential/parallel orchestration using git-spice and worktrees
Initialize spectacular environment - validate dependencies and configuration
Decompose feature spec into executable plan with automatic phase analysis
Generate a lean feature specification using brainstorming and the writing-specs skill
Use before starting any new phase - explains how sequential and parallel phases automatically chain together through base branch inheritance (main worktree tracks progress, parallel phases inherit from current branch, no manual intervention needed)
Use when working with stacked branches, managing dependent PRs/CRs, or uncertain about git-spice commands (stack vs upstack vs downstack) - provides command reference, workflow patterns, and common pitfalls for the git-spice CLI tool
Use when starting any conversation in a project using spectacular - establishes mandatory workflows for spec-anchored development, including when to use /spectacular commands and how to work with constitutions
Use to validate spectacular environment - checks superpowers plugin, git-spice, git repo, and project structure before running spectacular workflows
Use before creating worktrees or executing tasks - validates that CLAUDE.md defines required setup commands (install, optional postinstall) and provides clear error messages with examples if missing
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Spec-anchored development with automatic parallel execution for Claude Code.
Write a spec → generate a plan → execute tasks in parallel → get stacked PRs.
[!NOTE]
/spectacular:executerequires git-spice for PR stacking. Support for Graphite coming soon!
Inspired and powered by Superpowers.
| Problem | Without | With Spectacular |
|---|---|---|
| Context drift | "What was that pattern?" → makes something up | Every task gets a fresh subagent with anchored context |
| Sequential execution | 40 min of serial tasks (even independent ones) | Parallel tasks in isolated worktrees |
| Unmergeable PRs | 47 files, +2,847 lines—nobody reviews this | Stacked PRs, one per task |
| Pattern hallucination | AI invents plausible-sounding patterns | Constitutions define the rules |
Prerequisites:
# Install git-spice (macOS)
brew install git-spice
See git-spice installation instructions for other platforms.
Install plugins:
# In Claude Code, install marketplaces and plugins
/plugin marketplace add obra/superpowers-marketplace
/plugin install superpowers@superpowers-marketplace
/plugin marketplace add arittr/spectacular
/plugin install spectacular@spectacular
Initialize your project:
/spectacular:init
Real example from developing spectacular itself:
/spectacular:spec I can't use my main repo while spectacular is running, it locks everything up. I want each run isolated in its own worktree so I can keep working manually and run multiple features concurrently.
# Feature: Worktree Isolation for Spectacular Runs
## Problem Statement
Spectacular commands operate directly in the main repository:
- Cannot perform manual work during a spectacular run
- Cannot run multiple sessions concurrently
- No isolation between spectacular work and manual work
## Requirements
- FR1: Worktree creation in spec command
- FR2: Plan command operates in worktree
- FR3: Execute command uses worktree as base
...
## Acceptance Criteria
- [ ] Worktree created immediately after RUN_ID generation
- [ ] Spec committed to worktree branch
- [ ] Main repo working directory unchanged
...
/spectacular:plan @specs/dedf14-worktree-isolation/spec.md
## Execution Summary
- **Total Tasks**: 3
- **Total Phases**: 1
- **Sequential Time**: 12h
- **Parallel Time**: 5h
- **Time Savings**: 7h (58%)
## Phase 1: Command Worktree Integration
**Strategy**: parallel
**Reason**: All tasks modify different command files
### Task 1: Spec Command Worktree Integration
Files: commands/spec.md
Complexity: M (4h)
### Task 2: Plan Command Worktree Integration
Files: commands/plan.md
Complexity: M (3h)
### Task 3: Execute Command Worktree Integration
Files: commands/execute.md
Complexity: L (5h)
/spectacular:execute @specs/dedf14-worktree-isolation/plan.md
Time savings: 58% faster—parallel execution runs tasks simultaneously instead of sequentially.
Orchestrator
|
|-- 1. Read plan, identify 3 parallel tasks
|-- 2. Create worktrees
| dedf14-task-1/
| dedf14-task-2/
| dedf14-task-3/
|
|-- 3. Spawn subagents IN PARALLEL
| |
| +-- Subagent 1 (spec.md) +-- Subagent 2 (plan.md) +-- Subagent 3 (execute.md)
| | * Fresh spec context | * Fresh spec context | * Fresh spec context
| | * Implement, test | * Implement, test | * Implement, test
| | * Commit to branch | * Commit to branch | * Commit to branch
| | | |
| +-----------------------------+----------------------------+
| |
|-- 4. Collect results, stack branches: task-1 -> task-2 -> task-3
|-- 5. Code review, cleanup worktrees
gs stack submit
Result: 3 focused PRs instead of one large diff.
Working memory system for Claude Code - integrates with Obsidian via MCP for persistent, cross-session memory management
npx claudepluginhub joshuarweaver/cascade-code-general-misc-3 --plugin arittr-spectacularSpec Driven Development toolkit - structured specification, planning, and implementation workflows for systematic feature development
Skills-first specification-driven development framework with 7 agent skills for planning, implementation, review, and shipping. Natural language activation with intelligent agent orchestration. Includes /plan, /implement, /research commands plus managing-specifications, implementing-features, and reviewing-and-shipping skills.
Specification-Driven Development with Process Discipline for AI Agents
Spec-driven development using GitHub spec-kit methodology. Constitution-first approach with specify, plan, tasks, and implement phases.
Helder's personal SDLC toolbelt for AI coding agents — from spec to ship. Bundles the tracer-bullet workflow alongside TDD, code review, audits, and shipping skills.
A spec-driven development kit for Claude Code: the prompt-grill and dev-lifecycle skills plus a specs/ scaffold that takes you from a vague request to shipped, reviewed code.