Building an MCP-first tool: 7 design lessons from a shared memory for AI agents

Description and use

<p>Every agent session starts with amnesia. You explain the project, the conventions, the decision you made last Tuesday. The agent does good work, the session ends, and the next one starts from zero. Switch from Claude to Cursor to ChatGPT and you do it again, three times over.</p> <p>I got tired of pasting the same context into every chat, so I built <a href="https://korpus.cloud" rel="noopener noreferrer">Korpus</a>: one space where humans and their agents read and write the same memory. And I built it MCP-first. The MCP server is the main interface, the web app is a second client on the sa

Problem

<p>Every agent session starts with amnesia. You explain the project, the conventions, the decision you made last Tuesday. The agent does good work, the session ends, and the next one starts from zero. Switch from Claude to Cursor to ChatGPT and you do it again, three times over.</p> <p>I got tired of pasting the same context into every chat, so I built <a href="https://korpus.cloud" rel="noopener noreferrer">Korpus</a>: one space where humans and their agents read and write the same memory. And I built it MCP-first. The MCP server is the main interface, the web app is a second client on the sa

Outcome

<p>Every agent session starts with amnesia. You explain the project, the conventions, the decision you made last Tuesday. The agent does good work, the session ends, and the next one starts from zero. Switch from Claude to Cursor to ChatGPT and you do it again, three times over.</p> <p>I got tired of pasting the same context into every chat, so I built <a href="https://korpus.cloud" rel="noopener noreferrer">Korpus</a>: one space where humans and their agents read and write the same memory. And I built it MCP-first. The MCP server is the main interface, the web app is a second client on the sa

Evidence

Building an MCP-first tool: 7 design lessons from a shared memory for AI agents

Source