Tell one agent. Watch the team appear.
Implementation, independent Claude review and checks. Watch the agents appear and return their results to the coordinator.
Watch it work.
19 short films, recorded in the real app.
One agent summons ten. All ten report back.
Follow a real team through its live hierarchy, individual jobs and returning results.
A screenshot on your phone. In your agent’s hands.
Send a phone image directly to a coding agent, then bring in another agent to inspect the same image.
All your coding agents. One phone.
Move between different agent types on one phone and send each instruction to the right desktop conversation.
Let feedback become tomorrow’s work.
A scheduled agent triages feedback. Approved work flows through development, independent review and an updated outcome.
Wake up to an organised day.
Schedule an agent to gather your captured notes and organise your day in the shared notepad.
Change it on your phone. See it on your desktop.
Capture a note on your phone. Ask an agent to organise the list and see the result on both devices.
Say it. Send it. Watch your agent act.
Dictate on your phone, review the words and send them to your agent. Continue by voice on the desktop.
Hover a file. Open it. Keep the conversation.
Open a file from your agent’s output, follow its links and send the right path into the conversation.
Your agent writes a document. It appears beside the work.
A new Markdown document arrives in the sidebar. Open its formatted view without losing your agent’s place.
Switch projects. Keep the whole workspace together.
Work across repositories, switch workspaces and return to the right agents, files and notes.
Turn mixed changes into clear commits.
Ask an agent to group a mixed diff into meaningful commits, then inspect what belongs in each.
Start your first agent team.
Open a workspace and launch real agents from the native picker.
Give parallel tasks their own space.
Run independent tasks in separate working copies and branches.
Keep terminals, tests and saved tasks together.
Run a server, check the work and save a command for next time inside Code Overlord.
Put browser feedback in your agent’s hands.
Bring a real screenshot into the agent’s conversation, make the change and verify it.
Connect the tools your agents need.
Connect a documentation service in Code Overlord, then ask your agent to retrieve a real guide through it.
Leave a session. Return to the work.
Resume a real conversation and continue from its saved context.
Ask your agent to arrange the workspace.
Open a report and update the shared notepad with an instruction in the agent’s chat.
GuideGive each specialist its own role and accounts.
Two specialists use their own instructions and connected accounts, then bring their findings together.
GuideTake an issue through to a draft pull request.
Coordinate implementation and review, then follow the resulting pull request and checks inside the client.
GuideSend an agent to your cloud machine.
Connect Skyforge, launch a remote session and follow its real output from Code Overlord.
GuideInstall. Choose an agent. Begin.
Set up Code Overlord, authenticate a supported agent and start a working session.