Build with the TUI, CLI and SDK

​

Coming soon

• One agent, every client. Start in the dashboard, continue in the terminal app, and see the same session in both.
• Use the terminal app and the SDK. Run your own tools against your server with the OpenCode terminal app and the generated SDK.
• A server for each project or team. Run separate agent servers in one Workspace, so use and access stay apart.
• Suspend, resume and delete a server. Stop a server when you are done, bring it back later, and delete it when it is no longer needed.
• A readable address for your server. Give your server an address your team recognizes.
• A sandbox when the work needs one. The agent starts an execution sandbox for the first file or shell command. Questions that need no machine start none.
• Prepare one environment, fork it for many. Set up one environment, snapshot it and fork it for several agents.
• Open the sandbox yourself. Open the terminal and the running preview of what the agent built.
• A team of agents with roles. One agent plans, several implement in parallel and one reviews, all in one view, with your approval on the merge.
• Jobs on a schedule. Run a recurring job such as weekly dependency updates, with approvals waiting for you.
• Your own agent profiles. Draft a profile such as a security reviewer, approve it once, and let every agent in the Workspace use it.
• Run your own connector code. Deploy connector code that signs requests but never sees the secret.
• Use your own OAuth app. Connect through your own Google or Slack app, so Akua’s shared apps never touch your data.
• Continue in an existing execution sandbox. Pick up yesterday’s work with the same files, dependencies, and development server in your execution sandbox.
• Open your server at its own address. Use your agent server’s web app in a separate browser tab at its default address.
• Work together in one agent chat. Prompt one agent alongside a teammate and see messages and results live. Each person’s turn uses that person’s rights.
• Share a default execution sandbox and switch when needed. Continue a teammate’s work in the shared default execution sandbox. Create, fork, or switch environments in the same chat and see which agents use each one.
• Share an execution sandbox or keep it private. Share an execution sandbox with your team or keep it for yourself. Your agents see the same environments you can access.
• Build in your Workspace and ship to the team’s. Develop a Package with your agent in your own Workspace and install the result in your team’s production Workspace.
• Keep an agent server for selected people. Give one person or group access to a server so its chats, answers, and execution sandboxes stay within that group.
• Keep the sender’s rights throughout a task. Let a long task finish with the rights of the person who started it. A teammate repeating the request uses their own rights.
• Send approvals to the responsible person. Route an action from your turn to you or an admin named by your Workspace rules for approval.
• Stop or correct a teammate’s running turn. Stop a running turn or send a correction without waiting. Your correction starts a new turn with your rights, and queued messages wait until the chat is idle.
• Run tasks and apps in your Cluster. Ask the agent to start a temporary pod, schedule a CronJob, or deploy an app in a Cluster Akua manages, using your rights on that Cluster.
• Deploy from GitHub on every merge. Ask the agent to connect your GitHub repository and set up a pipeline that deploys to your Cluster on every merge.
• Run execution sandboxes in your own Cluster. Give agents execution sandboxes in your company’s Kubernetes Cluster through the same sandbox API.

​

Related topics