Connect Your IDE — Setup Guide

Tentra connects to your IDE via the Model Context Protocol (MCP). Zero install — just add a URL.

Step 1: Sign in and get your API key

Sign in with GitHub at trytentra.com/login, then go to Settings to generate your API key.

Step 2: Add to your IDE

Cursor: Settings → Features → MCP → Add Server. Claude Code: Add to .mcp.json in your project root.

Step 3: Create your first architecture

Describe your system in natural language — "Design a payment system with Stripe, Kafka, and PostgreSQL." Your AI assistant will call create_architecture and return a link to the visual diagram on Tentra's interactive canvas.

Step 4: Export production code

Ask your agent to "Export that architecture as a Spring Boot project" (or FastAPI, Go chi, Rust Axum, Ruby Rails — 14 frameworks total). The export_architecture tool produces a real project scaffold you can unzip and run.

Step 5: Index your repo (optional secondary feature)

Tell your AI to "Index this codebase with Tentra". The tool walks your files, extracts symbols and imports locally via Tree-sitter, then the agent annotates each file with purpose and domain tags. Future AI sessions query the persistent graph for secondary code-graph queries (symbols, call graphs, architectural hotspots) and drift detection against your saved architecture.

Test the connection

In your IDE chat, type: "List my architectures". If you see a response from the list_architectures tool, you are all set.

Full MCP Tools Reference

Tentra exposes 36 MCP tools — architecture workspace (including mid-task spec-alignment), code-graph write-path, code-graph read-path (including find_references and safe_rename for refactor safety), and enrichment (contracts, decisions, ownership, domains). Your AI assistant will automatically choose the right tool based on your prompt.

Troubleshooting