Nevios for developers

Nevios is a multi-tenant workspace platform — notes, kanban boards, tables, files, and inbox. These docs are for people who want to drive it programmatically — from their own scripts, AI agents, or third-party integrations.

Three ways to drive it

REST API — versioned endpoints under https://api.nevios.io/api/v1/. The same surface that powers the dashboard, just authenticated with a Personal Access Token instead of a session cookie.

MCP server — the same surface wrapped as AI-friendly tools. Plug it into Claude Desktop, Cursor, or Claude.ai and Nevios shows up in their tool menu. See MCP server.

Webhooks — coming soon. Push events to your own endpoint, HMAC-signed.

Architecture in a sentence

Nevios runs as a modular monolith (FastAPI + Postgres + Next.js). Every module — notes, sheets, files, inbox — ships as a self-contained package with its own models and routes.

Authorization is resource-level role based: viewer / editor / admin per folder, table, or file. The same checks gate the API and MCP — when an AI agent creates a note through MCP, it goes through the exact same auth pipeline as the web app. No backdoor, no elevated permissions.

What /v1 promises

• URLs are stable. Endpoints in /api/v1/ won't be renamed, removed, or change their response shape without a Sunset header and at least 90 days notice.
• Additive changes are fair game. New fields in a response, new endpoints, new scopes — any time. Your code should ignore fields it doesn't recognise.
• Breaking changes go under a new version. When we need to break, we mount under /api/v2/. v1 stays live until an explicit sunset.