About maind

maind is the context layer for AI-native development. We believe AI agents perform best when they have access to curated, current, codebase-specific lessons — not just whatever a base model happened to memorize.

Why we built it

Most AI tools assume the agent already knows your repo. In practice, agents reach for outdated patterns, miss your conventions, and drift between sessions. maind solves this by exposing a single MCP endpoint every agent already speaks — Claude Code, Claude Desktop, Cursor, Codex — backed by a lesson library that travels with your codebase.

What we ship

• MCP server. A streamable HTTP endpoint your agent can call to pull lessons, briefings, and structured context.
• Dashboard. Manage API keys, configure consent tiers, and watch how agents are using your context budget.
• Open lesson library. Community-contributed lessons for the Community tier; curated and reviewed lessons for paid tiers.

How we think about data

We run on Supabase EU-West and offer a three-tier consent model so you decide how much telemetry leaves your environment. Read the full breakdown in our Privacy Policy.

Get in touch

Questions, feedback, or partnership ideas: support@maind.dev.