Quick Start
Get Buildforce running in your project and complete your first spec-driven workflow in minutes.
Installation
Install the CLI
Install the @buildforce/cli package globally:
npm install -g @buildforce/cli
Or use npx without installing:
npx @buildforce/cli init .
Initialize in Your Project
Initialize Buildforce in an existing project:
buildforce init .
Or create a new project:
buildforce init my-project
Choose Your AI Assistant
During initialization, select your AI assistant:
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- And 7 more...
Buildforce will install slash command files into your assistant's configuration folder (e.g., .cursor/commands/).
Your First Workflow (Hello Buildforce)
Open your AI assistant in any existing project and run this workflow:
/research the architecture and structure of this codebase
/spec Update README.md to fix any inconsistencies with actual project structure and features
/build
/complete
This workflow works on any codebase! Here's what happens:
/research - Gather Context
Buildforce searches your context repository, explores your codebase, and produces a structured report with:
- File paths and their relevance
- Architecture diagrams
- Data models
- Key findings and recommendations
Research findings stay in context to inform your spec.
/spec - Define Requirements
Converts your description into structured requirements:
- Creates
spec.yaml(WHAT to build) with functional requirements, acceptance criteria, and scope - Creates
plan.yaml(HOW to build) with architecture decisions, phases, and tasks - Asks clarifying questions if anything is ambiguous
/build - Execute Implementation
Follows the plan systematically:
- Checks off tasks as work progresses
- Logs deviations from the original approach
- Validates against requirements
- Provides testing guidance
/complete - Finalize and Save Knowledge
Validates everything is done:
- Confirms all requirements met
- Reviews deviation log
- Generates context files capturing key decisions
- Updates context repository for future work
.buildforce/context/ that will inform future work. Each completed feature makes your context repository smarter.What Just Happened?
- Context loaded first -
/researchgathered existing context before requirements - Intent materialized -
/specconverted your request into structured requirements - Plan generated - Implementation approach defined with phases and tasks
- Work tracked - Progress and deviations logged transparently
- Knowledge saved - Decisions captured for future features
This is context-first spec-driven development in action.
Next Steps
Now that you've seen the workflow, dive deeper:
- Workflow Overview - Understand the workflow visualization
- Commands - Learn each command in detail
- Supported AI Assistants - See all supported assistants
/researchBefore building any feature, run /research [your-topic] to load relevant context. This ensures your specs are informed by existing patterns and past decisions.