Zero disk usage per project. Auto-updates on every run. Setup in 30 seconds.

Quick Start

Claude Code

claude mcp add inove-ai -- npx -y @joelbonito/mcp-server

Cursor / VS Code

Add to .cursor/mcp.json or .vscode/mcp.json:

{
  "mcpServers": {
    "inove-ai": {
      "command": "npx",
      "args": ["-y", "@joelbonito/mcp-server"]
    }
  }
}

Windsurf / Cline

Any MCP-compatible tool works — point it to npx -y @joelbonito/mcp-server as a stdio server.

What You Get

8 Tools

| Tool | Description | |------|-------------| | list_agents | List all 22 specialized agents | | list_skills | List all 41 modular skills | | list_workflows | List all 25 workflows | | get_agent | Get full agent content by name | | get_skill | Get skill content (SKILL.md + sub-files) | | get_workflow | Get workflow content by name | | route_task | Recommend agents + skills for a task description | | search_content | Full-text search across all content |

6 Resources

| URI | Description | |-----|-------------| | inove://agents | JSON list of all agents | | inove://agents/{name} | Full agent markdown | | inove://skills/{name} | Full skill markdown | | inove://workflows/{name} | Full workflow markdown | | inove://architecture | Architecture documentation | | inove://instructions | Framework instructions |

22 Prompts

One prompt per workflow: define, debug, create, brainstorm, enhance, deploy, test, track, status, log, finish, orchestrate, plan, preview, ui-ux-pro-max, review, test-book, release, squad, context, readiness, journeys.

Agents

| Agent | Domain | |-------|--------| | orchestrator | Multi-agent coordination | | project-planner | Architecture, system design | | product-manager | Requirements, discovery | | product-owner | User stories, backlog, MVP | | frontend-specialist | React, UI/UX, Tailwind | | backend-specialist | APIs, Node.js, server | | database-architect | Schemas, queries, migrations | | mobile-developer | iOS, Android, React Native | | security-auditor | Auth, OWASP, compliance | | penetration-tester | Offensive security testing | | debugger | Root cause analysis | | devops-engineer | CI/CD, Docker, infra | | test-engineer | Test strategies, TDD | | qa-automation-engineer | E2E, Playwright, automation | | performance-optimizer | Speed, Core Web Vitals | | seo-specialist | SEO, visibility, ranking | | documentation-writer | Technical documentation | | code-archaeologist | Legacy code, refactoring | | game-developer | Game logic, engines | | ux-researcher | User flows, wireframes, usability | | explorer-agent | Codebase analysis |

Usage Examples

User: "Help me debug this React authentication bug"
AI:   [calls route_task] → recommends debugger + security-auditor
      [calls get_agent("debugger")] → loads persona + protocols
      → Starts systematic debugging with root cause analysis

User: "List available workflows"
AI:   [calls list_workflows] → 25 workflows with descriptions
      → Shows /define, /debug, /create, /deploy, etc.

Requirements

• Node.js >= 22
• Any MCP-compatible AI tool

Development

git clone https://github.com/JoelBonito/inove-ai-framework.git
cd inove-ai-framework/mcp-server
npm install
npm run dev      # Reads from .agents/ filesystem (hot reload)
npm test         # 91 tests
npm run build    # Bundles content + compiles TypeScript

Architecture

• stdio transport for local use via npx
• Cloudflare Workers for remote SSE access
• Content from .agents/ is embedded at build-time into registry.ts
• In dev mode, reads directly from the filesystem for instant feedback
• Only 3 runtime dependencies: @modelcontextprotocol/sdk, yaml, zod

License

MIT