Skip to content

Feature Overview

A map of what Jaade does. Each card below links to a detailed page.

Running and reviewing agents

  • Agents — Claude Code, Codex, OpenCode, and Cursor, each with selectable models, reasoning effort, permission modes, and sandboxing.
  • Sessions & Chat — the New Task flow, the chat experience, the Session Overview canvas, pinned messages, connectors, and resumable history.
  • Work Summary — usage analytics: sessions, token curves, contributions, and agent activity comparisons.

The workspace

  • Terminal — real PTY shells in splittable panes, with themes, search, and clickable commit hashes.
  • Files & Editor — file explorer, an editable file viewer with markdown/diff preview, vim mode, LSP diagnostics, and rich previews.
  • Source Control & Git — diffs, intentional staging, commits, history, worktrees, conflict resolution, and AI-generated PRs.

Coordinating work

  • Projects — group sessions; a master session orchestrates workers through a task dependency graph.
  • Task Board — a kanban board synced from Jira, GitHub, and Linear, with per-task git isolation. (Experimental.)
  • Scheduled Tasks — repeatable prompts on manual, cron, or interval triggers.
  • Mini Apps — build and run small, git-versioned apps inside Jaade, optionally wired to the agent. (Experimental.)

Extending agents

  • Skills, Plugins & MCP — manage Claude plugins, Codex plugins, Agent Skills, and MCP servers from one screen.
  • Computer Use — let agents operate your macOS desktop. (Experimental, macOS.)
  • Browser Use — let agents drive your real Chrome. (Experimental.)
  • Memory — distil durable facts from sessions and inject them into future runs.

Helpers and remotes

  • SSH Remote Development — remote terminals and SFTP file browsing. (Experimental.)
  • Jaade Bot — a built-in assistant that explains the app and changes settings for you in plain language.

Configuration & reference

Released under the terms in the repository.