mait.sh1.142.0

Private beta: features and content may change. No warranty; use at your own risk.

MCP AIOM

Gives your AI agent the AIOM framework, its rules and its step-by-step skills to read on demand, so it works by your project's rules instead of guessing them.

Start here: what it does, why use it, and how it fits.

Beta
What it does

Your agent reads the project's rules instead of guessing them.

MCP AIOM lets your agent read the rules your workspace is meant to follow. AIOM is the MAIT framework for how a workspace is structured, documented, committed and delivered. The agent can see what exists, read exactly the piece it needs, check that your workspace follows the same framework version, and load step-by-step instructions for tasks such as moving to a newer version.

It only reads. The default collection is included with the application and needs no network; an owner may instead select an approved directory, and it never writes to your workspace or runs anything. What the agent does with the guidance is a separate step, under your workspace's own rules.

Your agentRules helperBuilt-in orapprovedrules1asks forrules2only readsYour agentRules helperBuilt-in or approved rules1asks forrules2only reads
  1. Step 1: Your agent asks for rules Rules helper.
  2. Step 2: Rules helper only reads Built-in or approved rules.
Why use it

What you gain over pasting rules into every conversation.

Rules that live in a chat are forgotten by the next session. Rules the agent can fetch, by name and version, remain available for each session to read; the agent must still apply them correctly.

Reliability

Rules fetched by name

The agent can consult the exact written rule

Typical alternative
You paste the conventions into each conversation
Here
The agent lists what exists and reads the exact rule it needs
Tokens

Only what the task needs

The conversation holds the relevant rules, not all of them

Typical alternative
A long rulebook is loaded whole, every time
Here
The agent reads the framework first, then only the rules that apply, and step-by-step instructions only when a task calls for them
Reliability

A version check

The agent reports the drift before it applies rules that no longer match

Typical alternative
Nobody notices the workspace is on an old framework
Here
The agent compares the framework version your workspace declares with the one it serves, and on a mismatch suggests the migration instructions without running them
Usability

Long instructions, in order

The agent can finish the complete instructions by following continuations

Typical alternative
A long instruction file arrives in one block
Here
Long instructions are handed over in parts, in order, so nothing is cut off silently
Speed

Read-only and offline

The default collection needs no network, and retrieval changes no workspace files

Typical alternative
Guidance fetched from a web page that can change under you
Here
The rules are built into the application, and the server can only read them
Experience

Your own collection

A team can give its agents its own approved rules in the same way

Typical alternative
One fixed set of rules for everyone
Here
Point the server at a directory your team approved and it serves that instead
Out of the box

Included in the image, one entry to connect.

MCP AIOM comes in the MAIT image (mait-image) and updates with it. Add one entry to your agent client and it starts when needed. The built-in collection works as it is, with no setup and no network.

You can also connect it through MCP Registry, so a single entry reaches all your tool servers. Get started shows how.

Where it fits

Where it sits in the MAIT toolkit.

Installed bymait-image ships the service
Reached throughMCP Registry, so one client entry reaches it
Used withMCP Git, for repository operations after the agent has read applicable Git guidance
Learn more

Where to read on.