# Concierge

## Docs

- [Code](https://docs.getconcierge.app/backends/code.md): Agent writes Python to call tools:maximum efficiency and composability
- [Overview](https://docs.getconcierge.app/backends/index.md): How Concierge exposes tools to the LLM and why the backend matters
- [Plain](https://docs.getconcierge.app/backends/plain.md): Expose all tools directly to the LLM with no wrapping
- [Plan](https://docs.getconcierge.app/backends/plan.md): Agent submits a JSON execution plan with sequential steps
- [Search](https://docs.getconcierge.app/backends/search.md): Semantic search over tools:discover then call
- [Installation](https://docs.getconcierge.app/getting-started/installation.md): Install Concierge and set up your development environment
- [Quickstart](https://docs.getconcierge.app/getting-started/quickstart.md): Build and run your first Concierge server in 5 minutes
- [Welcome](https://docs.getconcierge.app/home/index.md): Production grade MCP framework with progressive tool disclosure
- [Understanding MCP](https://docs.getconcierge.app/home/mcp101.md): What is MCP, why it matters, and what you can build with it
- [What is Concierge?](https://docs.getconcierge.app/home/what-is-concierge.md): Build efficient MCP servers with progressive tool disclosure, staged workflows, and code-based execution
- [Analytics](https://docs.getconcierge.app/platform/analytics.md): Monitor throughput, error rates, popular tools, and usage patterns
- [Evaluation](https://docs.getconcierge.app/platform/evaluation.md): Auto-evaluate your MCP server with mock user interactions and detailed reports
- [Inspector](https://docs.getconcierge.app/platform/inspector.md): Debug and test your MCP server with a visual interface
- [Logs](https://docs.getconcierge.app/platform/logs.md): View logs from your deployed servers
- [SDK Overview](https://docs.getconcierge.app/sdk/index.md): Stages, transitions, state, tools, resources, and widgets
- [Resources](https://docs.getconcierge.app/sdk/resources.md): Expose read-only data to AI agents via MCP
- [State](https://docs.getconcierge.app/sdk/state.md): Persistent per-session state with backend options
- [Tools](https://docs.getconcierge.app/sdk/tools.md): Define and expose functions to AI agents via MCP
- [Transitions](https://docs.getconcierge.app/sdk/transitions.md): Progressive tool disclosure via stages and transitions
- [Widgets](https://docs.getconcierge.app/sdk/widgets.md): Interactive UI components rendered inside AI conversations