Claude Code Setup Guide
Complete guide to integrating Bootspring with Claude Code (Anthropic's CLI coding assistant).
Overview
Claude Code is Anthropic's terminal-based AI coding assistant. Bootspring enhances Claude Code by providing:
- Structured Context — CLAUDE.md gives Claude full project understanding
- MCP Integration — 40+ tools accessible directly in your session
- Expert Agents — 44 specialized technical and business domain experts
- Build Loop — Task-driven development with
build next/build done - Observer Intelligence — Always-on session analytics and cost tracking
- Autopilot — Prescriptive recommendations injected in real time
Prerequisites
- Claude Code installed (
npm install -g @anthropic-ai/claude-code) - Bootspring CLI installed (
npm install -g @girardmedia/bootspring) - Node.js 18+
Quick Setup (Recommended)
The fastest path — one command configures everything:
bootspring setup writes the MCP entry to ~/.claude/settings.json automatically. No manual JSON editing required.
bootspring go runs the full workflow: detect stack → seed templates → merge context → generate AI files → init build loop.
What bootspring setup Configures
Running bootspring setup adds a bootspring MCP server entry to:
| Assistant | Config File |
|---|---|
| Claude Code | ~/.claude/settings.json |
| Codex | ~/.codex/config.toml |
| Gemini CLI | ~/.gemini/settings.json |
The entry looks like this in ~/.claude/settings.json:
Verify Setup
Add Ecosystem MCP Servers
Install popular third-party MCP servers alongside Bootspring:
Currently configures Context7 (library docs) and Sequential Thinking (step-by-step reasoning).
MCP Tools Available
When configured as an MCP server, Claude Code gets access to 40+ tools:
Core Tools
| Tool | Description |
|---|---|
bootspring_assist | Natural language assistant with recommendations |
bootspring_context | Query and manage project context |
bootspring_agent | Invoke specialized expert agents (44 types) |
bootspring_skill | Search and use 100+ code patterns |
bootspring_generate | Generate or regenerate CLAUDE.md |
Task & Automation Tools
| Tool | Description |
|---|---|
bootspring_todo | Manage development tasks |
bootspring_prd | Product requirement documents |
bootspring_memory | Access project memory and learnings |
bootspring_orchestrator | Workflow orchestration and phases |
bootspring_build | Build loop control (next/done/status) |
bootspring_seed | Project seeding and scaffolding |
bootspring_quality | Quality gates and checks |
Observer Intelligence Tools
| Tool | Description |
|---|---|
observer-session-metrics | Current session token/cost/tool metrics |
observer-cost-tracker | Cost tracking across time windows |
observer-summary | Aggregate stats and tool rankings |
observer-intelligence | Full 12-module intelligence report |
observer-efficiency | 6-dimension efficiency scoring (0–100) |
observer-patterns | Anti-pattern detection (thrashing, retry storms, etc.) |
observer-alerts | Smart alerts with dynamic thresholds |
observer-trends | Time series analysis |
observer-compare | Side-by-side session comparison |
observer-optimize | Cost optimization recommendations |
observer-replay | Session timeline replay |
Autopilot Intelligence Tools
| Tool | Description |
|---|---|
autopilot-status | Autopilot mode and feature status |
autopilot-config | Autopilot configuration |
autopilot-suggestions | Get prescriptive suggestions |
autopilot-learning | Suggestion effectiveness data |
System Tools
| Tool | Description |
|---|---|
bootspring_plugin | Plugin management |
bootspring_capabilities | Discover available capabilities |
bootspring_dashboard | Dashboard control and status |
bootspring_telemetry | Usage telemetry management |
bootspring_content | Content generation |
Using MCP Tools
Most of the time, Claude Code chooses the right tool automatically:
> Help me add a new API endpoint following our patterns
→ Claude uses bootspring_skill + bootspring_context
> Review this code for security issues
→ Claude uses bootspring_agent with security-expert
> What's my token usage this session?
→ Claude uses observer-session-metrics
You can also invoke tools explicitly:
> Use bootspring_agent database-expert to design the schema
> Use observer-efficiency to check my session score
> Use autopilot-suggestions to see what I should do next
Using Bootspring with Claude Code
Context-Aware Assistance
Claude Code reads your CLAUDE.md automatically:
Build Loop
The daily development workflow:
Invoking Agents
Ask Claude to use Bootspring agents:
> Use the bootspring security-expert to review my auth implementation
> Use the bootspring frontend-expert to build a dashboard component
> Use the bootspring database-expert to optimize these queries
Applying Skills
Request code patterns:
> Apply the bootspring stripe subscription pattern
> Search bootspring skills for authentication
Running Workflows
Start multi-phase workflows:
> Start the bootspring feature-development workflow for adding notifications
Configuration Files
CLAUDE.md (Auto-Generated)
Generated by bootspring generate — provides project understanding to Claude Code:
bootspring.config.js
Workflow Examples
New Project (One Command)
Existing Codebase
Feature Development
Troubleshooting
MCP Not Connecting
Context Not Loading
Agent Invocation Failures
Security
API Key Management
Bootspring uses browser-based device flow authentication — no API key management needed for standard usage.
For CI/CD or automated environments:
Sensitive Data
Exclude sensitive files from context generation:
Related
bootspring go— One-command project setupbootspring setup— MCP configurationbootspring doctor— Diagnose issues- MCP Tools Reference — All 40+ tools
- CLI Essentials — Full command reference