JaadeAgentic IDE
// Claude Code workspace

Run Claude Code
without losing the workspace.

Run Claude Code in a local macOS workspace with visible sessions, model and permission controls, terminals, files, diffs, history, plugins, skills, and MCPs.

Download for Mac
Public releases
// The working definition

Built around the run,
not just the editor.

Jaade wraps Claude Code in a local desktop workspace. The agent remains Claude Code; Jaade adds visible sessions, model and permission controls, files, terminals, diffs, history, schedules, plugins, skills, MCP status, and resumable context around each run.

Claude Code session in Jaade with workspace files, model controls, permissions, and changed-file contextClaude Code session in Jaade with workspace files, model controls, permissions, and changed-file context
Claude Code sessionA Claude Code run stays attached to its workspace, branch, controls, transcript, files, and review surfaces.
// What stays in reach

One surface for the work
around the agent.

Authentication

Use your Claude setup

Sign in with Claude, use standard API environment variables, configure AWS Bedrock, or point Claude Code at a supported compatible provider.

Control

Choose how the run starts

Set the Claude model, reasoning effort, permission mode, and plan mode before the first prompt, with attachments and workspace mentions in reach.

Extensions

See the capability surface

Browse Claude plugins, Agent Skills, MCP servers, authentication state, tools, and project scope before they become part of the session.

// A visible loop

Claude Code, with the run visible.

Jaade does not hide Claude Code behind an autonomous black box. It keeps the session, workspace context, approvals, tool activity, changed files, and follow-up controls visible throughout the run.

  1. 01

    Authenticate Claude

    Use Claude sign-in, your existing API environment, AWS Bedrock, or a supported Claude-compatible endpoint.

  2. 02

    Configure the session

    Select the model, reasoning effort, permission mode, and plan mode, then add files or attachments to the prompt.

  3. 03

    Steer and approve

    Watch the transcript and tool activity, answer inline permission prompts, and queue follow-ups while Claude is working.

  4. 04

    Review and resume

    Inspect the resulting diffs, commit intentionally, or reopen the same Claude session later with its original context intact.

// Continue with the details

Read the workflow, not a pitch.

View public repository
Agent setup

Configure Claude authentication, defaults, Bedrock, compatible providers, models, and context behavior.

Read the guide
Sessions

See how prompts, approvals, tool calls, attachments, history, and resume work inside a session.

Read the guide
Plugins and MCP

Manage Claude plugins, skills, commands, MCP servers, scopes, and authentication state.

Read the guide
// Questions

Before you
open a workspace.

Does Jaade include a Claude subscription or API access?

No. Jaade provides the desktop workspace and session controls. You authenticate Claude Code with your own Claude subscription, API configuration, Bedrock environment, or supported provider.

Can I use Claude Code plugins and MCP servers?

Yes. Jaade exposes Claude plugins, Agent Skills, commands, and MCP servers in a dedicated capability view, including scopes, enabled state, tools, and authentication status.

Is the Claude Code workspace local?

Jaade is a local macOS desktop app. Workspace files, session history, and app data stay on your machine, while prompts and tool calls still reach whichever agent provider or external service you configure.

Open a workspace.
Keep every session visible.

Jaade is a local Agentic IDE for Claude Code, Codex, OpenCode, and Cursor. Run agents and terminals, coordinate projects, schedule prompts, review diffs, and keep history, memory, plugins, skills, MCPs, and summaries in one workspace.

Download for Mac
Read the docs
Local desktop app for macOS Four coding agents Review, projects, memory, and schedules
local-first·agent-aware