Claude Code · Codex · MechaBee web

Bring your AI agent. MechaBee gives it your team's apps and files.

Install the MechaBee plugins in Claude Code or Codex to build your team's apps and work in them, right from the conversation. Or use our specialist agents for marketing and compliance in the browser. Every agent works in the same versioned, secure workspace as your team.

Sign in with your MechaBee account when your agent asks.

  • Claude Code

    with the MechaBee plugins

  • Codex

    with the MechaBee plugins

  • MechaBee web

    Marketor · DORA Compliance

One workspace

  • Apps

    records, boards, workflows

  • Files

    drafts, plans, assets

  • Your team

    same apps, same files

VersionedYour permissions
Install

Two plugins, one sign-in

Add the MechaBee marketplace and install both plugins. Your client asks you to sign in to MechaBee the first time an agent uses them.

Keeper App AuthorBuild new apps and maintain the ones you run.

Keeper Workspace ContentWork in your apps and files alongside your team.

Claude Code

Claude Code install
claude plugin marketplace add MechaBee/keeper-authoring-plugins
claude plugin install keeper-app-author@mechabee
claude plugin install keeper-workspace-content@mechabee

Then start a new Claude Code session so the plugins are loaded.

Updating? Run claude plugin marketplace update mechabee first, then the install line again.

Codex

Codex install
codex plugin marketplace add MechaBee/keeper-authoring-plugins
codex plugin add keeper-app-author@mechabee
codex plugin add keeper-workspace-content@mechabee

Then start a new Codex task so the plugins are loaded.

Updating? Run codex plugin marketplace upgrade mechabee first, then the install line again.

Your agent asks you before every change it writes. New to MechaBee? Create your account first, or read the plugin details.

With the plugins

What your agent can do in MechaBee

From the first version of an app to the daily work inside it, from the same conversation.

01

Build a new app

Describe how your team works: what you track, who does what, a normal week. Your agent builds a real app with records, boards, dashboards and workflows, live on desktop and mobile.

  • Personal trackers or shared team records
  • Workflows and in-app agent actions that work on your data
  • Open it with one click when it's live

You ask

Build a service dispatch board with customer and equipment records and a parts list.

Your agent builds

CustomersEquipmentPartsDispatch boardJobs workflow
02

Keep it running as your work changes

Ask for a change, and your agent shows you the before and after. Nothing goes live until you approve it, and existing records are kept by default.

  • Every change checked against the live app before you see it
  • Import data, and move to bigger storage as the app grows
  • Repair a deployment, or retire an app you no longer need
Describe
Review
Approve
Live

Add an approval step before jobs close

+ Jobs: “Waiting for approval” status

+ Approve job action for managers

~ Board: new column before Done

Records: 1,284 kept

03

Share the work with your team

Your team works in the app. Your agent works in the same app: it picks up a ticket, posts updates as it goes, and reports back where the team will see it. You decide which steps wait for a person to approve.

  • Reads the queues and records an app shares with agents
  • Saves its work through the app's own steps, checked before they're saved
  • Everything it writes is signed “Agent (on behalf of …)”, naming the person who ran it

Invoice PDF shows the wrong VAT total WEB#142

  • Agent (on behalf of Sam Rivera)

    Update. Started on the VAT total fix. Moved to In progress.

  • Agent (on behalf of Sam Rivera)

    Fixed. Invoice PDFs now total VAT per line. Closed as completed.

Jamie Chen was notified.

Agent Projects · a Keeper app

Where the work stands shouldn't be buried in a session

Every agent session knows how far it got, and none of the others can see it. Switch machines, accounts, or from Claude Code to Codex, and you either find that exact session again or explain the project from scratch. Agent Projects keeps the project's progress outside any one session, so the next session or teammate starts where the work actually stands.

Agent Projects project page with the next step and the history of session checkpoints

Checkpoints from any session

Ask your agent to push progress: what was done and the one next step. The newest push sets what the project shows; history keeps the rest.

A living handoff

One handoff per project: where it stands, the backlog, decisions and how to check the work. A fresh session reads it first.

Recognised by its work

Projects remember their repository, folders, plans and key files, so an agent finds the right project, along with how to run it.

Shared and planned

Share projects with your team by domain, see due dates and dependencies on a timeline, and get a ranked priorities review.

Agent Projects timeline with projects grouped by domain and dependency arrows
In the browser

Or use our specialist agents

No agent of your own? MechaBee's agents run in the web app, each specialised for one job, and they work in the same workspace as everything else.

Marketor

Marketing that ships every week

A marketing team's weekly routine: strategy, blog posts, social content, a campaign calendar, images and WordPress publishing, all kept in your workspace.

See Marketor

DORA ComplianceEarly access

Continuity documentation for DORA

Runs the discovery interviews with your CTO, CEO, operations and legal leads, then drafts your business continuity and disaster recovery plans, ICT vendor register and testing schedule, and checks them against DORA's requirements.

Ask for early access

Every web agent knows your workspace: mention a file by name and it reads it, and what you make together is saved back into the project.

Underneath

One workspace for every agent and every app

Claude Code, Codex and MechaBee's own agents all work in the same place your team does.

Versioned

Every change is a revision you can inspect and restore, so letting an agent edit isn't a gamble.

No silent overwrites

An agent's edit is checked against the version it read. If someone changed the file since, it stops instead of overwriting.

Your permissions, not the agent's

Every call is limited to what the signed-in person can reach, and apps add record-level access on top.

Apps and files together

An app built in Claude Code is the same app your team opens in the browser, next to the same files.

Start with the agent you already use

Install the plugins in Claude Code or Codex, or start in the browser with our own agents.