mex gives your whole team one shared memory layer for AI coding agents. Validated against your real code and kept in sync across every developer.
The open-source mex CLI is live today. Try it on GitHub
Each engineer's AI agent learns the codebase its own way, so the team quietly splits onto different paths. Memory fragments across machines, drifts from the real code, and no one sees the gap until it ships as inconsistent code.
01 / Fragments
Each developer's agent learns the codebase on its own, so no two share the same memory. Two agents solve the same problem different ways, and the conflict only surfaces in review, after it's already in the PR.
02 / Goes stale
The codebase changes every day; the per-developer memory does not. Agents keep enforcing rules that no longer match what's shipped, and nothing flags the gap.
03 / Ungoverned
There's no shared source of truth and no way for a lead to review, correct, or govern the memory every agent is working from. The team's standards end up living in tools leadership can't see into.
Every developer and every agent works from one shared, server-hosted memory. A team admin governs what gets promoted, and each developer still keeps their own local context.
Admin dashboard
governs sync, drift, access, and policy
shared · server-hosted
Individual context, received per dev.
See every agent's drift score at a glance, review what gets promoted into shared memory, and keep the whole team's context honest, without leaving the dashboard.
mex is vendor-neutral and works with the agents your team already uses, with a dashboard so leads can keep shared context honest as the team grows.
mex is open source and moving fast. Here's what's new in the CLI, and how we think about memory for AI coding teams.
Makes the code graph affordable to keep. Stores are roughly 40% smaller, graph builds hold one compiler program at a time, and interrupted builds can be repaired using mex graph repair.
Makes the code graph useful as a first-response retrieval system. mex graph scope now returns the most relevant source declarations and execution flows under a hard token budget.
Adds a deterministic code knowledge graph beneath the existing markdown scaffold. Memory can now ground itself to exact code nodes using mex graph ground.
by mex · on mex
On a set of retrieval tasks, mex returned answers that were 10x more compact than a grep-based baseline, while still finding everything it needed to find.
Read moreby mex · on LinkedIn
Every few months, a new model arrives with a larger context window. But being present is not the same as being used.
Read on LinkedInby Daksh Jaitly · on X
Why every session starts cold, and the case for persistent, structured memory over giant context files.
Read on XWe started like most teams: every developer keeping their own rules file, agents drifting apart within weeks. Corrections piled up, context went stale, and no one had a clear view of what our AI tools actually knew.
So we built mex: shared and validated memory for the whole team, governed in one place. AI-assisted development should get better as your team grows, not harder.
From scattered local rules to one shared source of truth.
