Package Exports
This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (spec-flow) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.
Readme
An open toolkit that turns product ideas into production launches through Spec-Driven Development.
🆕 Recent Updates
v3.2.0 (November 2025)
Planning Optimization - 50-60% faster feature planning with docs/project integration
- Mandatory Project Docs Integration:
/plannow parsesdocs/project/files for instant context- Tech stack (tech-stack.md): Framework, backend, database, styling, state management
- Data architecture (data-architecture.md): Existing entities, naming conventions
- API strategy (api-strategy.md): API style, auth provider, versioning, error format
- Performance targets (capacity-planning.md): Scale tier, response time targets
- Architecture (system-architecture.md): Monolith/microservices/serverless
- Deployment (deployment-strategy.md): Platform and model
- Research Phase Optimization: 5-15 tools → 2-5 tools (with project docs)
- Eliminates: Codebase scanning for stack detection, API pattern inference, entity discovery
- Preserves: Component reuse scanning, feature-specific research
- Freshness Validation: Compares docs to code, warns if mismatch detected
- Enhanced research.md: Project Context section shows parsed values from all 8 docs
- Time Savings: 4-8 minutes per feature, 3-4 hours per project (30 features)
Problem Solved: Before v3.2.0, /plan re-researched information already documented in docs/project/, leading to slower planning and potential tech stack hallucination. No integration with /init-project output.
Impact: Planning phase is 50-60% faster (8-15 min → 4-7 min). Zero tech stack hallucination. Consistent architecture across all features. Clear attribution of decisions to source docs. Better designs that respect performance targets and scale tier.
v3.1.0 (November 2025)
OKLCH Color Space Upgrade - Perceptually uniform design tokens
- OKLCH Color Space: Upgraded
/init-brand-tokensto generate OKLCH colors- 92% browser support with automatic sRGB fallback
- Perceptually uniform lightness across all hues
- Wide gamut support (P3 color space, 50% more colors than sRGB)
- Semantic Token Structure: Replaced DEFAULT/light/dark with explicit bg/fg/border/icon tokens
- Eliminates "ink vs paint" confusion
- Automated WCAG AAA (7:1) contrast validation using colorjs.io
- New Token Categories: Focus (WCAG 2.2), Motion (reduced-motion), Data viz (Okabe-Ito colorblind-safe), Typography features, Dark mode shadows (3-6x opacity)
- Design Principles: Comprehensive OKLCH documentation and best practices
- Added dependency: colorjs.io ^0.5.0 for OKLCH conversion and WCAG validation
Problem Solved: v1.0.0 used RGB colors with false WCAG claims (e.g., claimed 7.8:1 but actually 5.70:1). No semantic token structure caused "ink vs paint" confusion. Missing critical tokens for focus, motion, and accessibility.
Impact: Perceptually uniform colors with accurate WCAG validation. Clear semantic meaning for every token. WCAG 2.2 compliant focus indicators. Colorblind-safe data visualization. Proper dark mode depth perception.
v3.0.1 (November 2025)
Package Fix - Added missing design commands to npm
- Fixed npm package: Added missing design workflow commands
/init-brand-tokens- Smart brand token initialization/design- Unified design orchestrator (Variations → Functional → Polish)/research-design- Multi-source design inspiration gathering/update-project-config- Project configuration updates (was missing after rename)
- Removed deprecated: Removed old
setup-constitutionreference
Problem Solved: v2.12.0 introduced comprehensive design workflow commands, but they were accidentally excluded from the npm package files list. Users who installed via npm couldn't access these commands.
Impact: Design workflow is now fully available in npm package. Users can initialize brand tokens, run design workflows, and configure projects without cloning the repo.
v3.0.0 (November 2025)
Constitution Cleanup - Simplified project structure and documentation
- BREAKING CHANGE: Constitution file structure reorganized
- Split monolithic
constitution.mdinto 3 focused files:docs/project/engineering-principles.md- User-editable 8 core engineering principlesdocs/project/project-configuration.md- User-editable deployment model and scale tier settings.spec-flow/memory/workflow-mechanics.md- Workflow system mechanics (roadmap, versioning, quality gates)
- Split monolithic
- Enhanced /init-project: Now generates 10 comprehensive docs (was 8)
- Renamed Commands:
/setup-constitution→/update-project-config(76% size reduction: 862→207 lines) - Roadmap Integration:
/roadmap brainstormnow reads tech-stack.md and capacity-planning.md for technically grounded features - Greenfield Support: Automatic foundation issue creation for new projects
- Fixed Broken References: Updated all commands to reference new file structure
Problem Solved: Before v3.0.0, constitution.md was a monolithic 609-line file mixing user project settings with workflow system mechanics. Confusing dual-constitution naming (constitution.md vs setup-constitution.md). Roadmap brainstorming didn't consider technical constraints.
Impact: Clear separation of concerns - user files vs workflow files. Smaller, focused files easier to maintain. Foundation setup automated for greenfield projects. Features are technically validated from brainstorming phase.
v2.12.0 (November 2025)
S-Tier Design Workflow - Complete UI/UX design system without Figma
- Three-Phase Pipeline:
/designorchestrator chains Variations → Functional → Polish with human approval gates - Token-Based Design System: Smart initialization with brownfield/greenfield detection via
/init-brand-tokens - Automated Validation:
design-lint.jsenforces principles (elevation scale, 2:1 hierarchy, subtle gradients, token compliance) - Comprehensive Handoff:
implementation-spec.md(26KB) provides pixel-perfect component breakdown, interactions, accessibility, testing strategy - Usability Evaluation:
dont-make-me-think-checklist.mdwith 130 checkpoints across 10 categories (visual clarity, navigation, content, interactions, feedback, cognitive load, conventions, error prevention, mobile, accessibility) - Brownfield Scanner: Analyzes existing codebases to detect and consolidate design patterns (e.g., 47 colors → 12 tokens)
- Frontend Integration: Frontend-shipper agent validates implementation against design artifacts with automated design lint checks
Problem Solved: Before v2.12.0, achieving high-quality UI/UX required Figma and manual design handoff, leading to inconsistent implementations and design-dev drift. Manual validation of design principles (elevation, hierarchy, gradients) was error-prone.
Impact: Systematic enforcement of S-tier design principles from wireframes through implementation. 100% token compliance ensures consistent UI. Pixel-perfect handoff documentation eliminates design-dev miscommunication. Brownfield support enables design system adoption in existing projects.
v2.10.0 (October 2025)
X Announcement Slash Command - Converted X announcement from skill to slash command
- Direct Invocation:
/x-announce "version"replaces skill-based workflow - 5-Option Workflow: Post now, schedule, draft, edit, or skip
- Threaded Replies: Automatic GitHub release link as reply using
in_reply_to_tweet_id - Release Integration: Integrated into
/releasecommand Step 9 (optional, non-blocking) - Simplified Architecture: Command-based execution replaces skill loading
Problem Solved: Before v2.10.0, X announcements required loading a skill and complex interaction flow. Now /x-announce provides direct, streamlined access with same functionality.
Impact: Release announcements are faster and simpler. Command structure is more maintainable than skill-based approach.
v2.9.0 (October 2025)
Help Command - Context-aware workflow navigation
- Contextual Guidance:
/helpshows where you are and what to do next - Six Output Modes: Adapts to context (no feature, in feature, at gate, blocked, complete, corrupted)
- Progress Visualization: Emoji indicators show phase status (✅ ⏳ ⬜ ❌)
- Deployment Model Aware: Shows correct phase sequences for staging-prod, direct-prod, local-only
- Error Recovery: Highlights blockers with specific recovery options
- Verbose Mode:
/help verbosefor detailed state (quality gates, deployments, artifacts) - Always Actionable: Shows next command to run based on current state
Problem Solved: Before v2.9.0, users had to manually track workflow state and remember next steps. Now /help provides instant orientation and guidance.
Impact: Reduced cognitive load and faster error recovery. No more "what do I run next?" confusion.
v2.8.1 (October 2025)
npm Package Fix - Properly excludes internal workflow files
- Fixed npm package: Internal files (release.md, x-announcement.md) no longer included in published package
- Explicit file listing: package.json files array updated with explicit includes and negation patterns
- Enhanced protection: .npmignore created for additional safety layer
- Security: Prevents accidental exposure of internal API URLs and workflow development files
Problem Solved: Before v2.8.1, internal workflow development files leaked into npm package. Now proper exclusion prevents exposure.
Impact: Enhanced security - internal API URLs and development files stay private.
v2.8.0 (October 2025)
X Announcement Integration - Automated social media announcements for releases
- Automated X Posts:
/releasenow posts announcements to X (Twitter) automatically with custom post generation - Threaded Replies: GitHub release link posted as threaded reply using
in_reply_to_tweet_idAPI parameter - Preview & Edit: Review and customize post text before sending (280 character limit validation)
- Graceful Fallback: Manual posting instructions if X Poster API unavailable
- Security: X announcement skill and updated /release command gitignored (internal API URL protection)
Problem Solved: Before v2.8.0, release announcements required manual X posting. Now /release handles social media automatically with customizable posts and proper threading.
Impact: Release visibility increased with zero-friction social announcements. Internal API URLs remain protected.
v2.7.0 (October 2025)
Release Automation & Essential Finalization - One-command releases and universal branch cleanup
- Full /release Automation: CI validation, README updates, CHANGELOG, git tags, GitHub releases, and npm publishing in one command
- Smart Version Detection: Analyzes conventional commits to automatically determine MAJOR/MINOR/PATCH bumps
- Pre-flight Validation: Checks git remote, branch, working tree, npm auth, and CI status before release
- Auto-Issue Closing: GitHub issues automatically close with "shipped" status when features deploy
- Essential Finalization: Roadmap updates and branch cleanup now run for ALL deployment models (not just staging-prod)
- Clean CI: Fixed all ShellCheck warnings in bash scripts for reliable CI passes
Problem Solved: Before v2.7.0, releases required 10+ manual steps and direct-prod/local-only deployments skipped cleanup tasks. Now /release handles everything automatically and all deployment models clean up properly.
Impact: Release workflow is now 5x faster (2 minutes vs 10+ minutes) and branch cleanup happens consistently across all deployment models.
v2.6.0 (October 2025)
TodoWrite Progress Tracking - Clear visibility and error recovery for /ship and /finalize
- Visual Progress Tracking:
/shipnow tracks all 5-8 deployment phases with TodoWrite - Documentation Tasks Tracked:
/finalizetracks all 10 documentation tasks (CHANGELOG, README, releases, issues, branches) - No More Silent Failures: See exactly what's happening when CI errors occur
- Error Recovery: Specific "Fix [error]" todos added when builds/deployments fail
- Resumability: Run
/ship continueafter fixing errors to resume from failed phase - Manual Gate Clarity: Preview and staging validation gates clearly marked as pending
Problem Solved: Before v2.6.0, /ship would stop silently on CI errors and /finalize would skip tasks without visibility. Now users see current progress, blockers, and can resume after fixes.
Impact: Deployment workflows are now transparent and recoverable, preventing the frustration of "it stopped and I don't know why."
v2.9.0 (October 2025)
Help Command - Context-aware workflow navigation
- Contextual Guidance:
/helpshows where you are and what to do next - Six Output Modes: Adapts to context (no feature, in feature, at gate, blocked, complete, corrupted)
- Progress Visualization: Emoji indicators show phase status (✅ ⏳ ⬜ ❌)
- Deployment Model Aware: Shows correct phase sequences for staging-prod, direct-prod, local-only
- Error Recovery: Highlights blockers with specific recovery options
- Verbose Mode:
/help verbosefor detailed state (quality gates, deployments, artifacts) - Always Actionable: Shows next command to run based on current state
v2.8.1 (October 2025)
npm Package Fix - Properly excludes internal workflow files
- Fixed npm package: Internal files (release.md, x-announcement.md) no longer included in published package
- Explicit file listing: package.json files array updated with explicit includes and negation patterns
- Enhanced protection: .npmignore created for additional safety layer
- Security: Prevents accidental exposure of internal API URLs and workflow development files
v2.8.0 (October 2025)
X Announcement Integration - Automated social media announcements
- Automated X Posts:
/releaseposts to X automatically with threaded GitHub link reply - Preview & Edit: 280 character validation with preview before posting
- Custom Post Generation: Extracts CHANGELOG highlights for engaging posts
- Graceful Fallback: Manual posting instructions if X Poster API unavailable
- Security: Internal files gitignored to protect API URLs
v2.1.2 (October 2025)
/feature next Auto-Pull - Automatically start highest priority roadmap item
- Usage:
/feature next- Auto-pulls highest priority feature from GitHub Issues - Smart Search: Queries
status:nextfirst, falls back tostatus:backlog - Priority Sorting: Sorts by
priority:high→priority:medium→priority:low - ICE Display: Shows Impact/Confidence/Effort scores and calculated ICE score
- Auto-Status Update: Moves issue from
next/backlog→in-progress - Full Traceability: Links workflow state to GitHub issue number
- No Manual Input: Just run
/feature next- no feature description needed
Benefits: Work on highest priority items automatically, reduce context switching, maintain roadmap sync.
v2.1.0 (October 2025)
GitHub Issues Roadmap Migration - Complete migration from markdown to GitHub Issues
- Backend Migration:
/roadmapcommand now uses GitHub Issues instead of markdown files - Label-Based State: Features organized via labels (
status:backlog,status:next,status:in-progress,status:shipped) - ICE Scoring Preserved: Impact × Confidence / Effort stored in YAML frontmatter in issue descriptions
- Priority Labels: Auto-applied based on ICE score (
priority:high>= 1.5,priority:medium0.8-1.5,priority:low< 0.8) - Dynamic Sorting: Features sorted by priority labels via queries (no manual file editing)
- Native Integration: Issues, PRs, and roadmap in one place with GitHub Projects support
- Cross-Platform: Both Bash and PowerShell examples throughout
- Same Workflow: All actions (add, brainstorm, move, delete, ship) work identically
Breaking Change: Roadmap data is now in GitHub Issues, not .spec-flow/memory/roadmap.md. Old markdown roadmap can be migrated using .spec-flow/scripts/bash/migrate-roadmap-to-github.sh.
Setup Required: Run .spec-flow/scripts/bash/setup-github-labels.sh to create labels, then authenticate with gh auth login or set GITHUB_TOKEN.
Documentation: See docs/github-roadmap-migration.md for complete guide.
v1.13.0 (October 2025)
Local Project Integration Workflow - Automatic merge-to-main for local-only projects
- New Phase S.4.5a: Merges feature branch → main/master after successful local build
- Auto-detects main branch: Supports both
mainandmasterbranch names - Preserves feature history: Uses
--no-ffmerge to maintain branch context - Remote sync: Automatically pushes to origin if remote exists
- Branch cleanup: Offers to delete feature branch locally and remotely after merge
- Correct sequencing: Merge happens BEFORE version bump and roadmap update
- Git best practices: Version tag created on main branch (not feature branch)
Before: optimize → build-local → finalize ❌ (stayed on feature branch)
After: optimize → build-local → merge-to-main → finalize ✅ (integrated to main)
Impact: Local-only projects now have complete parity with remote deployment models - all features properly integrate to main branch before being marked "Shipped" in roadmap.
v1.12.1 (October 2025)
Update Command Simplification - Removed backup overhead for faster, cleaner updates
- Removed backup creation:
npx spec-flow updateno longer creates backup folders - Faster updates: No backup overhead, instant template updates
- User data still preserved: Templates updated while learnings.md, memory, and specs remain untouched
- Removed --force flag: No longer needed (backwards compatible)
- Cleaner output: Shows "Templates updated, user data preserved" message
Why? The preserveMemory flag already protects user data during updates. Backups created redundant folders that users had to manually clean up. This change simplifies the update process while maintaining safety.
v1.12.0 (October 2025)
Learnings Persistence & Design Iteration - Skills learn across npm updates + enhanced design workflow
Part 1: Learnings Persistence (All 16 Skills)
- Separated learnings data from SKILL templates: Created
learnings.mdfor all 16 phase skills - Preserves knowledge across npm updates: SKILL.md templates get updated, learnings.md data persists
- Tracks pitfall frequencies: Auto-updates frequency stars (⭐☆☆☆☆ → ⭐⭐⭐☆☆) as issues occur
- Tracks pattern success rates: Records usage counts and success rates for proven approaches
- Zero manual intervention: System learns automatically as you work
Part 2: Design Iteration Enhancements
- Screen-specific targeting:
/design-variations $SLUG [$SCREEN]- iterate on single component - Overwrite protection: Warns before regenerating, offers [b]ackup to create git tag first
- Re-enable support: Can enable design workflow after initially declining in
/feature - Iteration patterns guide: 5 common scenarios with step-by-step solutions (component iteration, A/B testing, state-specific refinement)
Documentation
- Added "Skills & Learning System" section to README
- Two-file architecture explained (templates vs data)
- Iteration patterns guide for design workflow
Files: 36 changed (16 learnings.md created, 16 SKILL.md updated, iteration-patterns.md created)
v1.11.1 (October 2025)
Folder Cleanup Refactor - Cleaner spec directories with on-demand folder creation
- Eliminated empty folders: Removed blanket directory pre-creation from
/speccommand - On-demand creation: Folders (
visuals/,design/) now created only when files are written - Benefits: Cleaner spec directories, easier to identify UI features, follows YAGNI principle
- Files modified:
.claude/commands/specify.md
v1.11.0 (October 2025)
Phase-Specific Learning Skills - Workflow improvement tracking
- Added learning Skills for each workflow phase (spec, plan, tasks, implement, optimize, ship, etc.)
- Auto-triggers on workflow events to capture lessons and improve future iterations
- Skills track common issues: clarification overload, test failures, deployment blockers
Table of Contents
- What is Spec-Driven Development?
- Get Started
- Supported AI Agents
- Script Reference
- Core Philosophy
- Development Phases
- Prerequisites
- Learn More
- Detailed Process
- Troubleshooting
- Packages & Releases
- Maintainers
- License
🌟 Why Spec-Flow?
Building software with AI assistants is powerful, but without structure, projects drift. You lose context, forget decisions, skip testing, and ship inconsistent features. Spec-Flow solves this.
The Problem Without Spec-Flow
| Challenge | Without Spec-Flow | With Spec-Flow |
|---|---|---|
| Context Loss | "What were we building again?" after interruptions | NOTES.md tracks all decisions, checkpoints restore context instantly |
| Inconsistent Quality | Features shipped without tests, reviews vary | Every feature follows same rigorous process: spec → plan → implement → review → ship |
| Token Waste | Conversations balloon to 100k+ tokens, Claude slows down | Auto-compaction at 80% budget keeps context efficient (75k/100k/125k per phase) |
| No Accountability | "Did we test this? Who approved?" | Auditable artifacts for every phase, approval gates enforced |
| Reinventing Process | Each feature starts from scratch | Reusable templates, proven patterns, documented workflows |
What You Get
✅ Repeatable Process - Every feature follows the same proven workflow (spec → plan → tasks → ship)
✅ Context Discipline - Token budgets enforced per phase, auto-compaction prevents context bloat
✅ Quality Gates - Automated checks for accessibility, performance, testing, security
✅ Auditable Trail - Every decision documented in NOTES.md, every phase produces artifacts
✅ Faster Velocity - Skip decision paralysis, let the workflow guide you
✅ Team Alignment - Specs reviewed upfront, parallel work enabled, consistent outcomes
Use Cases
- Web Apps - Full-stack features with frontend + backend coordination
- APIs - Contract-first development with automated testing
- CLIs - Command structure definition to distribution
- Mobile Apps - Offline-first architecture with platform-specific handling
- Design Systems - Component libraries with accessibility built-in
- Infrastructure - Terraform modules with security scanning
- ML Projects - Experiment tracking with reproducible pipelines
👉 See more: Use Cases
What is Spec-Driven Development?
Spec-Driven Development flips the traditional model: specifications become executable assets that orchestrate planning, implementation, QA, and release. Each Claude command owns a phase of delivery, produces auditable artifacts, and tees up the next specialist.
The Workflow
💡 Ideas → 🗺️ Roadmap → 📝 Spec → 📋 Plan → ✅ Tasks → 🔍 Analyze →
💻 Implement → ⚡ Optimize → 👀 Preview → 🚀 Staging → ✅ Validate → 🎉 ProductionKey Principle: Plan your roadmap first, then write specifications from prioritized features, and let AI agents execute faithfully.
🚀 New: Optimized Phase Agent Architecture (v1.5.0)
Spec-Flow now features an optimized orchestrator (/feature) that runs each workflow phase in isolated contexts for maximum efficiency:
Benefits:
- ⚡ 67% Token Reduction - Each phase runs in isolated context (240k → 80k tokens per feature)
- 🏃 2-3x Faster - No cumulative context bloat, no /compact overhead between phases
- ✅ Same Quality - All slash commands unchanged, proven workflow maintained
How it works:
/feature (Orchestrator - Lightweight State Tracking)
├→ spec-phase-agent → /spec → Returns summary
├→ plan-phase-agent → /plan → Returns summary
├→ tasks-phase-agent → /tasks → Returns summary
├→ /implement → Spawns worker agents directly (bypasses phase agent*)
└→ ... (each phase isolated, efficient handoffs)
* Phase 4 calls /implement directly due to sub-agent spawning limitsChoose your workflow:
/feature "feature"- Full feature workflow (recommended)/quick "fix"- Fast path for small changes (<100 LOC)
🚀 Quick Start
Option 1: NPM (Recommended)
Install Spec-Flow with a single command:
# Initialize in current directory
npx spec-flow init
# Or\spec-flow target directory
npx spec-flow init --target ./my-projectOption 2: Manual Installation
Clone and run the interactive wizard:
# 1. Clone Spec-Flow repository
git clone https://github.com/marcusgoll/Spec-Flow.git
cd Spec-Flow
# 2. Run the installation wizard (Windows)
powershell -File .spec-flow/scripts/powershell/install-wizard.ps1
# OR (macOS/Linux)
./.spec-flow/scripts/bash/install-wizard.shWhat gets installed:
- ✅
.claude/- Agents, commands, and settings - ✅
.spec-flow/- Scripts, templates, and memory - ✅
CLAUDE.md- Workflow documentation - ✅
QUICKSTART.md- Quick start guide (copied to your project) - ✅ Memory files initialized with defaults
Next steps after installation:
- Read the guide - Open
QUICKSTART.mdin your project - Open in Claude Code - Navigate to your project directory
- Set up your project (optional but recommended):
/setup-constitution # Interactive Q&A for engineering standards /roadmap # Plan and prioritize features with ICE scoring /design-inspiration # Curate visual references for consistency
- Start building:
/feature "my-feature" # Full feature workflow /quick "fix bug" # Fast path for small changes
👉 Full guide: QUICKSTART.md | Detailed tutorial: Getting Started
Get Started
1. Install the toolkit
From npm (fastest):
npm install -g spec-flow
# or use npx without a global install
npx spec-flow init --target ./my-projectFrom source:
Clone this repository and ensure you have either PowerShell 7.3+ (pwsh) or a POSIX shell (bash) plus Python 3.10+ available. Scripts live under .spec-flow/scripts/powershell/ and .spec-flow/scripts/bash/.
Full installation guide: docs/installation.md
Copy .claude/settings.example.json to .claude/settings.local.json and update the allow list for your environment.
2. Establish principles
Run the /setup-constitution command in Claude to document the engineering principles that guard every feature. Store the output in .spec-flow/memory/setup-constitution.md.
3. Build your roadmap
Use /roadmap to manage features via GitHub Issues with ICE scoring (Impact × Confidence / Effort):
- Setup: Authenticate with
gh auth loginor setGITHUB_TOKEN, then run.spec-flow/scripts/bash/setup-github-labels.sh - Add features:
/roadmap add "feature description"creates GitHub issue with priority labels - Brainstorm:
/roadmap brainstorm [quick|deep]generates ideas from research - Organize via labels:
status:backlog- Ideas to considerstatus:next- Top 5-10 prioritized featuresstatus:in-progress- Currently being builtstatus:shipped- Completed features (issue closed)
- View roadmap: Browse GitHub Issues or use
gh issue list --label status:next
Documentation: See docs/github-roadmap-migration.md for complete guide.
4. Kick off a feature
Select a feature from your roadmap and choose your workflow:
Full workflow (recommended):
/feature "feature-name" # Runs full workflow with isolated phase agents
# Auto-progresses through: spec → plan → tasks → validate → implement → optimize → ship
# Pauses at manual gates: /preview, /validate-staging
# Use: /feature continue (to resume after manual gates)Manual step-by-step:
/spec "feature-name" # Create specification
/plan # Create plan
/tasks # Break down tasks
/implement # Execute implementation
# ... continue through remaining phasesSupported AI Agents
| Agent | Status | Notes |
|---|---|---|
| Claude Code | Supported | Optimised for slash-command workflow. |
| Cursor | Supported | Pair with .spec-flow/memory/ context files. |
| Windsurf | Supported | Share roadmap + constitution for planning. |
| GitHub Copilot | Partial | Works for code edits; manual command execution. |
| Gemini CLI | Experimental | Requires manual prompt translation. |
Skills & Learning System
Spec-Flow includes an auto-learning system that captures lessons from each workflow phase to continuously improve your process.
How Skills Work
16 Phase-Specific Skills: Each workflow phase has a dedicated skill that learns from execution:
.claude/skills/specification-phase/- Learn from /spec (reduce clarifications, improve classification).claude/skills/planning-phase/- Learn from /plan (maximize code reuse, detect patterns).claude/skills/implementation-phase/- Learn from /implement (TDD enforcement, anti-duplication).claude/skills/ui-ux-design/- Learn from design workflow (Jobs principles, a11y compliance)- ...and 12 more for other phases
Two-File Architecture:
- SKILL.md - Template with pitfall descriptions, detection logic, prevention strategies (updated with npm)
- learnings.md - Accumulated data with frequencies, metrics, instances (preserved across updates)
Learnings Persistence
When you update Spec-Flow via npm, your accumulated learnings are preserved:
What Gets Updated (npm overwrites):
SKILL.md- Template improvements, new pitfalls, better detection logic.claude/commands/*.md- Command improvements, new features.spec-flow/templates/*.md- Artifact templates
What Gets Preserved (your local data):
learnings.md- Pitfall frequencies, pattern usage counts, metrics- GitHub Issues - Your product roadmap (stored as issues with labels)
.spec-flow/memory/setup-constitution.md- Your project principlesspecs/*/- All your feature specifications
Example: After 10 features, your learnings.md shows:
common_pitfalls:
- id: "over-clarification"
frequency: 2/5 ⭐⭐☆☆☆
last_seen: "2025-01-15"
instances: [feature-001, feature-007]
successful_patterns:
- id: "informed-guess-strategy"
usage_count: 8
success_rate: 87.5%This data persists across npm updates, so your workflow gets smarter over time.
Auto-Learning in Action
Skills auto-trigger when:
- Starting a phase (loads relevant lessons)
- Completing a phase (detects pitfalls, updates frequencies)
- Encountering errors (captures patterns for prevention)
Skills auto-update when:
- Pitfall detected: Increments frequency (0/5 → 1/5 → 2/5 ★★☆☆☆)
- Pattern used successfully: Increments usage count, recalculates success rate
- Metrics updated: Averages recalculated after each feature
No manual intervention required - the system learns as you work.
Script Reference
Every automation script is provided in both PowerShell (.ps1) and shell (.sh) form. Pick the variant that matches your environment.
| Task | Windows / Cross-platform | macOS / Linux |
|---|---|---|
| Validate prerequisites | pwsh -File .spec-flow/scripts/powershell/check-prerequisites.ps1 -Json |
.spec-flow/scripts/bash/check-prerequisites.sh --json |
| Scaffold a feature | pwsh -File .spec-flow/scripts/powershell/create-new-feature.ps1 "Dashboard revamp" |
.spec-flow/scripts/bash/create-new-feature.sh "Dashboard revamp" |
| Estimate token budget | pwsh -File .spec-flow/scripts/powershell/calculate-tokens.ps1 -FeatureDir specs/015-dashboard |
.spec-flow/scripts/bash/calculate-tokens.sh --feature-dir specs/015-dashboard |
| Compact context | pwsh -File .spec-flow/scripts/powershell/compact-context.ps1 -FeatureDir specs/015-dashboard |
.spec-flow/scripts/bash/compact-context.sh --feature-dir specs/015-dashboard |
Additional scripts such as
enable-auto-merge,wait-for-ci, andupdate-agent-contextalso ship with.shwrappers that delegate to PowerShell so you can run them from a POSIX shell while we build native equivalents.
Core Philosophy
- Specification first every artifact traces back to an explicit requirement.
- Agents as teammates commands encode expectations so assistants stay aligned.
- Context discipline token budgets are measured, compacted, and recycled.
- Ship in stages staging and production have dedicated rituals with human gates.
Development Phases
| Phase | Command | Primary Outputs |
|---|---|---|
| -1 | /roadmap |
GitHub Issues with ICE-scored features and priority labels |
| 0 | /spec |
spec.md, NOTES.md, visuals/README.md |
| 0.5 | /clarify |
Clarification log inside the spec |
| 1 | /plan |
plan.md, research.md |
| 2 | /tasks |
tasks.md with acceptance criteria |
| 3 | /validate |
Risk analysis report |
| 4 | /implement |
Implementation checklist & validation hooks |
| 5 | /optimize |
Code review summary & optimization plan |
| 6 | /debug |
Error triage and remediation plan |
| 7 | /preview |
Release notes & preview checklist |
| 8 | /ship-staging |
Staging deployment ritual |
| 9 | /validate-staging |
Sign-off for staging |
| 10 | /ship-prod |
Production launch and follow-up |
| - | /compact [phase] |
Optional: Reduce token usage between phases |
Context Management: The /compact command is optional and reduces token usage by summarizing verbose artifacts. Use it between phases when context feels heavy or when suggested by auto-progression.
Prerequisites
- Git 2.39+
- Python 3.10+
- PowerShell 7.3+ (
pwsh) for Windows scripts - Bash 5+ (or Zsh) for shell scripts
- Claude Code access with slash-command support
- yq 4.0+ for YAML state management:
- macOS:
brew install yq - Linux: See mikefarah/yq releases
- Windows:
choco install yq
- macOS:
- PowerShell-yaml module for PowerShell scripts:
Install-Module -Name powershell-yaml -Scope CurrentUser
Optional:
- GitHub CLI (
gh) for auto-merge helpers - Pester 5 for PowerShell test suites
📚 Examples
Complete Working Example: Dark Mode Toggle
Explore a fully-documented feature workflow in specs/001-example-feature/:
specs/001-example-feature/
├── spec.md # Feature specification with user scenarios
├── NOTES.md # Progress tracking and decisions
├── artifacts/
│ ├── plan.md # Implementation plan with architecture
│ ├── tasks.md # 28 tasks with acceptance criteria
│ ├── analysis-report.md # Risk assessment (0 critical issues)
│ ├── optimization-report.md # Performance metrics (145ms avg)
│ └── release-notes.md # v1.2.0 release notes
└── visuals/
└── README.md # Design references and color tokensWhat's included:
- Complete specification with FR/NFR requirements
- 28 tasks broken down across 5 implementation phases
- Performance benchmarks (27% better than target)
- WCAG 2.1 AA accessibility compliance
- Cross-browser testing matrix
- Release notes for v1.2.0
👉 Browse the example: specs/001-example-feature/
Learn More
docs/architecture.md— how the repository fits togetherdocs/commands.md— quick lookup for every slash commanddocs/getting-started.md— step-by-step tutorial for your first featuredocs/installation.md— platform-specific installation guidedocs/troubleshooting.md— common issues and solutionsdocs/use-cases.md— examples for different project typesPUBLISHING.md— release checklist for npm & GitHub PackagesAGENTS.md— contributor guide for working in this repoCONTRIBUTING.md— branching, reviews, and release process
Detailed Process
- Run
.spec-flow/scripts/bash/check-prerequisites.sh --json(or the PowerShell variant) to ensure your environment is ready. - Set up roadmap: Authenticate with GitHub (
gh auth login), run.spec-flow/scripts/bash/setup-github-labels.shto create labels. - Build roadmap: Use
/roadmapto add features as GitHub Issues, prioritize with ICE scoring, organize via labels (status:backlog → status:next → status:in-progress → status:shipped). - Select a feature from GitHub Issues and launch
/feature "<feature-slug>"in Claude to scaffold the spec from the issue. - Progress through
/clarify,/plan,/tasks, and/validate, addressing blockers as they appear. - Use
calculate-tokensto watch context budgets andcompact-contextto summarise when approaching thresholds. - Walk the release staircase:
/preview,/ship-staging,/validate-staging,/ship-prod. - The feature is automatically marked as shipped in GitHub Issues (label changed to
status:shipped, issue closed), and changelog is updated with the release.
Packages & Releases
- npm: Published as
spec-flow. Install globally withnpm install -g spec-flowor run one-off withnpx spec-flow. - GitHub Packages: The
Publish Packagesworkflow mirrors each release to GitHub Packages under the scoped name@marcusgoll/spec-flow, enabling the repository's Packages tab. - Automation: Creating a GitHub release (or manually running the workflow) triggers the dual publish. Set the
NPM_TOKENrepository secret with an npm automation token that haspublishrights; GitHub packages use the built-inGITHUB_TOKEN.
Troubleshooting
| Issue | Resolution |
|---|---|
pwsh command not found |
Install PowerShell 7 (winget install Microsoft.PowerShell or brew install --cask powershell). |
| Shell script reports missing feature directory | Run /feature first or use create-new-feature to scaffold specs/NNN-slug. |
| Token estimate returns zero | Verify files are UTF-8 encoded and not empty. |
| Context delta lacks checkpoints | Ensure NOTES.md records checkpoints prefixed with -. |
Maintainers
- Marcus Gollahon (@marcusgoll)
- Community contributors join via pull requests!
License
Released under the MIT License.