Emily for Code  ·  Powered by MLA

Keep coding agents aligned with the decisions still in force.

Claude Code and Codex can retrieve the right information and still act on an obsolete decision. MLA maintains the active source of truth that keeps coding sessions aligned.

Claude Code and Codex ·  Apache-2.0  ·  Free while in beta

Same coordination system. Different intervention.

With people, Emily asks and follows up. With coding agents, MLA puts the current decision into the agent’s context before it acts.

How it works

  1. 1

    Remember what was decided

    MLA maintains the human-approved decisions still in force, so there is one answer to what is current rather than a pile of everything ever written.

  2. 2

    Catch conflicting assumptions

    When a coding session starts acting from stale or contradictory context, MLA identifies the mismatch instead of letting the session run on it.

  3. 3

    Put the current decision in front of the agent

    Claude Code or Codex gets the relevant current context before it proceeds, at the moment it would otherwise have guessed.

Set it up from your coding agent.

mla runs inside Claude Code and Codex rather than in your terminal. Paste one prompt into your agent and it installs mla and sets up the repo for you.

Open source  ·  Read the client on GitHub