Learn
Learn — the shared memory layer for AI coding agents.
Evergreen guides on AI agent memory, MCP, and spec-driven development — the concepts behind giving coding agents a durable, project-scoped source of truth.
What is AI Agent Memory?
The durable context an agent uses across tasks — and why coding agents need shared project memory, not private chat history.
What is Coding Agent Memory?
The project context a coding agent reads before it edits the repo: specs, architecture, decisions, plans, and execution history.
Long-Term Memory for Coding Agents
Durable project context that survives across sessions and runs, so each new agent run starts from the project's truth — not a cold prompt.
Shared Memory vs Agent Memory
Private agent memory personalizes one assistant; shared memory gives a team and its agents one project-scoped source of truth.
Project Memory Layer
The durable, project-scoped source of truth that sits between prompts and code, holding the design agents build from.
What is an MCP Server?
A service that exposes tools, resources, and prompts to AI agents over the Model Context Protocol — explained in plain developer terms.
MCP Memory Server
An MCP server that gives agents durable, project-scoped memory to read and update — not just one-off tools or chat history.
What is Spec-Driven Development?
Defining behavior, architecture, and acceptance criteria before implementation — and why it matters more when agents code fast.
Writing AI-Friendly Software Requirements
A practical guide to requirements coding agents can build from: explicit scope, constraints, acceptance criteria, and architecture context.
Give your agents a source of truth.
Put these ideas to work — a project-scoped memory layer your coding agents read and write over MCP.