Context Engine

People who use multiple AI tools such as Claude Code, Cursor and ChatGPT find that each one starts every session from nothing, forgetting decisions made the day before or in another tool. What one person's AI learns stays locked in that person's chat history, invisible to teammates. As a result, people and teams end up re-explaining the same project, rules and conventions in every tool and every session, and knowledge never accumulates anywhere the whole team can see.

Context Engine is a hosted service at app.contextengine.com that acts as one shared memory for a person's or team's AI tools. Tools connect once through MCP (ChatGPT and Claude via a few clicks, Claude Code via one command, Cursor via an add link) and then save memories such as decisions, notes, people, customers and files as they work, each tagged with type, source tool and timestamp. Memory is organized into 'engines', each with its own purpose statement, database, search index and access controls, so a client's or project's information stays walled off from others. The system writes its own daily/weekly summaries, links related memories, prevents accidental duplicates, keeps a full edit history, and lets people or tools query it in plain language with answers that cite their source memories. All data can be exported at any time as Markdown and JSON, and engines can be deleted (with a one-day export grace period).