Architecture Overview

Understanding how Bootspring works under the hood.

System Architecture#

Bootspring is a multi-component system that provides AI-assisted development tools:

┌─────────────────────────────────────────────────────────────────────┐ │ CLIENT LAYER │ ├─────────────────────────────────────────────────────────────────────┤ │ │ │ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ │ │ CLI │ │ Claude │ │ VS Code │ │Dashboard │ │ │ │ │ │ Desktop │ │Extension │ │ (Web) │ │ │ └────┬─────┘ └────┬─────┘ └────┬─────┘ └────┬─────┘ │ │ │ │ │ │ │ └────────┼───────────────┼───────────────┼───────────────┼──────────┘ │ │ │ │ │ │ MCP Protocol │ │ ▼ ▼ ▼ ▼ ┌─────────────────────────────────────────────────────────────────────┐ │ MCP SERVER │ ├─────────────────────────────────────────────────────────────────────┤ │ │ │ ┌──────────────────────────────────────────────────────────┐ │ │ │ Tool Registry │ │ │ │ assist | context | agent | skill | workflow | ... │ │ │ └──────────────────────────────────────────────────────────┘ │ │ │ │ │ ┌──────────────────────────┼──────────────────────────────┐ │ │ │ ▼ │ │ │ │ ┌────────────┐ ┌────────────┐ ┌────────────┐ │ │ │ │ │ Agent │ │ Skill │ │ Workflow │ │ │ │ │ │ Engine │ │ Engine │ │ Engine │ │ │ │ │ └────────────┘ └────────────┘ └────────────┘ │ │ │ │ Core Engines │ │ │ └──────────────────────────────────────────────────────────┘ │ │ │ └─────────────────────────────────────────────────────────────────────┘ │ │ │ Local │ Remote ▼ ▼ ┌─────────────────────────┐ ┌─────────────────────────┐ │ Local Storage │ │ Bootspring API │ ├─────────────────────────┤ ├─────────────────────────┤ │ │ │ │ │ .bootspring/ │ │ Authentication │ │ context/ │ │ Usage Tracking │ │ agents/ │ │ License Validation │ │ skills/ │ │ Cloud Sync │ │ config.json │ │ │ │ │ │ │ └─────────────────────────┘ └─────────────────────────┘

Core Components#

Client Layer#

End-user interfaces:

ComponentDescription
CLICommand-line tool for direct interaction
Claude DesktopMCP integration for Claude
VS Code ExtensionIDE integration
DashboardWeb-based management

MCP Server#

The Model Context Protocol server:

  • Exposes tools to AI assistants
  • Manages local state
  • Orchestrates workflows
  • Communicates with APIs

Core Engines#

Business logic components:

EnginePurpose
Agent EngineManages expert agent invocations
Skill EngineHandles pattern application
Workflow EngineOrchestrates multi-phase processes
Context EngineGenerates and manages context
Quality EngineRuns quality gates

Data Flow#

Agent Invocation Flow#

User → CLI/MCP → Agent Engine → Context Assembly → LLM → Response → User
  1. User requests agent invocation
  2. CLI/MCP receives request
  3. Agent engine loads agent profile
  4. Context engine assembles relevant context
  5. Request sent to LLM
  6. Response processed and returned

Context Generation Flow#

Codebase → Analyzers → Aggregator → Template → CLAUDE.md
  1. File system scanned
  2. Multiple analyzers extract information
  3. Results aggregated
  4. Template applied
  5. Context file generated

Workflow Orchestration#

Workflow Definition → Phase Manager → Agent Coordination → Completion Signals
  1. Workflow loaded from definition
  2. Phase manager tracks progress
  3. Agents invoked for each phase
  4. Completion signals validate progress

Architecture Docs#

DocumentDescription
Bootspring InternalsCore system design
MCP ProtocolMCP implementation details
Context SystemContext generation
Agent SystemAgent architecture
Workflow EngineWorkflow orchestration
Plugin SystemExtension architecture

Design Principles#

1. Local-First#

Bootspring operates primarily locally:

  • No data leaves your machine without consent
  • Works offline for most features
  • API calls only for auth/tracking

2. Protocol-Based#

Uses standard protocols:

  • MCP for AI integration
  • REST for API
  • JSON for configuration
  • Git for version control

3. Extensible#

Built for extension:

  • Plugin architecture
  • Custom agents
  • Custom skills
  • Custom workflows

4. Non-Invasive#

Doesn't modify your code unless asked:

  • Generates context files
  • Suggests changes via AI
  • Never auto-commits

Technology Stack#

CLI & MCP Server#

TechnologyPurpose
Node.jsRuntime
TypeScriptLanguage
Commander.jsCLI framework
MCP SDKProtocol implementation

API & Dashboard#

TechnologyPurpose
Next.jsWeb framework
PostgreSQLDatabase
PrismaORM
ClerkAuthentication
StripePayments
VercelHosting

AI Integration#

TechnologyPurpose
Claude 3.5Primary AI model
OpenAI (optional)Alternative provider
LangChain (internal)LLM orchestration

Deployment Architecture#

Local Installation#

User Machine ├── bootspring (CLI) ├── .bootspring/ │ ├── config.json │ ├── context/ │ └── cache/ └── Your Project ├── CLAUDE.md └── bootspring.config.js

Cloud Services#

┌─────────────────┐ │ Vercel CDN │ └────────┬────────┘ │ ┌──────────────┼──────────────┐ │ │ │ ▼ ▼ ▼ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ Dashboard │ │ API │ │ Webhooks │ │ (Next.js)│ │ (Next.js)│ │ Handlers │ └─────┬─────┘ └─────┬────┘ └─────┬────┘ │ │ │ └──────────────┼──────────────┘ │ ┌──────────────┼──────────────┐ │ │ │ ▼ ▼ ▼ ┌──────────┐ ┌──────────┐ ┌──────────┐ │PostgreSQL│ │ Redis │ │ S3 │ │ (Neon) │ │(Upstash) │ │(Backups) │ └──────────┘ └──────────┘ └──────────┘

Security Architecture#

Authentication Flow#

User → Clerk → JWT → API → Resource

API Key Security#

  • Keys hashed with bcrypt
  • Scoped permissions
  • Audit logging
  • Rate limiting

Data Protection#

  • TLS in transit
  • Encrypted at rest
  • No code stored server-side
  • GDPR compliant