Skip to content
Back to search

yume

Native desktop UI for Claude Code with orchestration, streaming, background agents, and multi-provider support. Built with Tauri + React.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/aofp/yume with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/aofp/yume'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/aofp--yume/badge.svg)](/listings/aofp--yume)

HTML

<a href="/listings/aofp--yume"><img src="/api/listings/aofp--yume/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 250 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # yume
  2. <div align="center">
  3. <img src="assets/yume.png" alt="yume logo" width="120">
  4. <h3>claude code, supercharged</h3>
  5. <p>orchestration flow. live interleaved thinking. background agents. plugins & skills.<br>4 built-in agents. vim mode. 4 providers. 7 languages. mac, windows, linux.</p>
  6. <p>
  7. <a href="https://aofp.github.io/yume">Download</a>
  8. </p>
  9. </div>
  10. ---
  11. ## why yume?
  12. Claude Code is transformative for development. Terminal interfaces hold it back.
  13. ### terminal limitations
  14. - **flickering** — terminal redraw artifacts, a long-standing cli complaint
  15. - **input lag** — terminal rendering slows down in extended sessions
  16. - **single session per window** — parallel work means juggling terminals
  17. - **rate limits on demand only** — requires manual `/usage` checks
  18. - **no gui affordances** — no visual diffs, clickable rollback, or drag-and-drop
  19. ### yume's approach
  20. - ✅ **native rendering** - hardware-accelerated UI, zero terminal artifacts
  21. - ✅ **responsive input** - typically <50ms, independent of session length
  22. - ✅ **standalone app** - your session doesn't die with your terminal or IDE
  23. - ✅ **persistent rate limit display** - 5-hour and 7-day metrics always visible
  24. - ✅ **automatic session recovery** - 5-minute snapshots with 24-hour retention
  25. ---
  26. ## how it works
  27. yume is **not** an alternate harness or api wrapper. we spawn the **official claude code cli** as a subprocess.
  28. ```
  29. yume app → spawns real cli binary → cli handles auth → anthropic servers
  30. ```
  31. this is fundamentally different from tools that spoof headers to bypass api pricing. yume:
  32. - runs the actual `claude` binary installed on your machine
  33. - uses your existing claude pro/max subscription
  34. - never touches anthropic's api directly
  35. - maintains 100% cli compatibility (subagents, mcp, hooks, skills, claude.md all work)
  36. think of it like **iterm vs terminal.app** - a different frontend for the same tool.
  37. ---
  38. ## features
  39. ### orchestration flow (unique to yume)
  40. - **automatic task decomposition** - understand → decompose → act → verify workflow
  41. - **complex task handling** - 3+ step tasks get architect planning, implementer execution, guardian verification
  42. - **trivial task passthrough** - simple tasks proceed directly without overhead
  43. - **no commands needed** - baked into every session via --append-system-prompt
  44. ### agent system
  45. - **4 built-in agents** - synced to ~/.claude/agents/yume-*.md
  46. - **architect** - plans, decomposes tasks, identifies risks, todowrite tracking
  47. - **explorer** - read-only codebase search, glob/grep/read, never modifies
  48. - **implementer** - focused code changes, edit/write, minimal diffs
  49. - **guardian** - reviews for bugs, security, performance after changes
  50. - **4 concurrent background agents** - async execution with git branch isolation (yume-async-{type}-{id})
  51. - **30-minute timeout per agent** - output to ~/.yume/agent-output/, 24h retention
  52. - **streaming isolation** - background agents don't interfere with main session
  53. - **merge conflict detection** - pre-merge checks before integrating agent work
  54. - **worktree isolation** - full filesystem isolation via git worktree
  55. - **agent coordination** - inter-agent messaging to hand off work between background agents
  56. ### multi-provider support
  57. - **15+ models, 4 providers** - non-claude providers via yume-cli shim
  58. - claude fable 5 (default flagship), sonnet 5 (1M context), sonnet 4.6, opus 4.8, haiku 4.5
  59. - gemini 3.1 pro preview, gemini 3 flash preview
  60. - gpt-5.5, gpt-5.4 mini
  61. - kiro (auto, claude sonnet 4.5, claude haiku 4.5, glm-5)
  62. - **protocol normalization** - stream-json output for all providers
  63. - **same interface** - switch providers without changing workflow
  64. ### core features
  65. - **persistent rate limit visibility** - 5h and 7d limits always visible, no /usage needed
  66. - **zero visual artifacts** - native rendering eliminates terminal flickering
  67. - **interleaved thinking toggle** - watch claude reason live, word-by-word, with on/off control
  68. - **keyboard-first** - 50+ shortcuts, full mouse-free navigation
  69. - **<50ms response time** - instant UI regardless of session length
  70. - **standalone reliability** - independent process, immune to IDE crashes
  71. - **auto-compaction** - dynamic threshold (default 85%), user configurable or disable
  72. - **effort + budget controls** - low/medium/high/xhigh effort and thinking budget per model
  73. - **usage-limit guard** - optionally stops new claude turns before the 5-hour window hits 100%, so a subscription never spills into metered overage
  74. - **schedule system** - `/schedule 5m`, `/schedule 2pm`, `/schedule done` for timed and event-based tasks
  75. - **askuser ipc** - claude asks structured questions mid-session with choice popover
  76. - **secret censoring** - api keys and tokens automatically redacted in all output
  77. - **7 languages** - english, spanish, french, german, polish, chinese, japanese
  78. ### plugin & skills system
  79. - **complete extensibility** - commands, agents, hooks, skills, MCP servers, monitors
  80. - **23 hook events** - pre-tool-use, post-tool-use, context warnings, compaction, subagent lifecycle, and more
  81. - **skills auto-injection** - context based on file extensions, keywords, regex
  82. - **plugin directory** - install/enable/disable per component
  83. - **bundled yume plugin** - default commands and agents
  84. ### additional features
  85. - **crash recovery** - 5-min snapshots, 24hr retention, complete session restore
  86. - **history rollback** - restore points per message, conflict detection, undo edits
  87. - **mid-stream context** - real-time token updates during streaming
  88. - **analytics dashboard** - usage by project/model/date, cost tracking, export
  89. - **voice dictation** - F5 to toggle, native speech-to-text
  90. - **@ mention system** - @ shows recent/modified files with autocomplete, folder navigation
  91. - **vim mode** - full normal/insert/visual/command modes for chat navigation
  92. - **row split panels** - up to 6 panels in 3x2 grid with F7/F8 split/combine
  93. - **rules management** - .claude/rules/ CRUD in settings
  94. - **image thumbnails** - drag-and-drop images with compression pipeline
  95. - **slash commands** - /schedule, /loop, /compact, /clear, /model, /resume, /usage, plus custom and cli commands
  96. ### customization
  97. - **18 themes** - 12 dark (OLED-optimized) + 6 light; accents fully customizable
  98. - **OLED optimization** - pure black (#000000) default
  99. - **100% CLI compatibility** - subagents, MCP, hooks, skills, claude.md, /commands
  100. ---
  101. ## tech stack
  102. - **tauri 2.10** - rust backend, native speed
  103. - **react 19** - modern ui with virtualized rendering
  104. - **sqlite + wal** - sessions, messages, analytics persistence
  105. - **socket.io** - real-time streaming with mid-stream context updates
  106. - **compiled binaries** - no node.js dependency for end users (@yao-pkg/pkg)
  107. - **3-process architecture** - tauri (rust) + react ui + node.js server
  108. - **zero telemetry** - your code stays on your machine
  109. ---
  110. ## comparison
  111. same cli, same claude subscription — different frontend. what yume adds on top of a plain terminal:
  112. | feature | cli in a terminal | cli in yume |
  113. |---------|-------------------|-------------|
  114. | interface | text ui, one session per window | native app — tabs, panes, windows |
  115. | orchestration flow | manual prompting | ✓ baked into every session |
  116. | background agents | run more terminals | ✓ 4 parallel, branch/worktree isolated |
  117. | thinking display | plain text scrollback | ✓ live stream with toggle + timers |
  118. | usage limits | `/usage` on demand | ✓ 5h + 7d always visible, overage guard |
  119. | context | `/context` on demand | ✓ live token bar, one-key compact |
  120. | file changes | text diffs | ✓ visual diffs + per-message rollback |
  121. | crash recovery | `--resume` after restart | ✓ 5-min snapshots, full layout restore |
  122. | providers | claude | ✓ claude + gemini + openai + kiro |
  123. | customization | terminal theme | ✓ 18 themes, custom colors, fonts, vim mode |
  124. | price | free with your claude plan | free demo · $4/mo · $49 lifetime |
  125. yume licenses the app — claude usage stays on your existing claude plan or api key, with no markup. ide-style tools (cursor, windsurf, …) solve a different problem: they bundle model access into $20+/mo subscriptions.
  126. ---
  127. ## download
  128. visit [aofp.github.io/yume](https://aofp.github.io/yume) for all download options.
  129. ---
  130. ## keyboard shortcuts
  131. | action | macOS | Windows/Linux |
  132. |--------|-------|---------------|
  133. | new tab | ⌘T | Ctrl+T |
  134. | close tab | ⌘W | Ctrl+W |
  135. | duplicate tab | ⌘D | Ctrl+D |
  136. | fork session | ⌘⇧D | Ctrl+Shift+D |
  137. | command palette | ⌘P | Ctrl+P |
  138. | files panel | ⌘E | Ctrl+E |
  139. | git panel | ⌘G | Ctrl+G |
  140. | model & tools | ⌘K | Ctrl+K |
  141. | toggle model | ⌘⇧K | Ctrl+Shift+K |
  142. | open project | ⌘O | Ctrl+O |
  143. | settings | ⌘, | Ctrl+, |
  144. | search messages | ⌘F | Ctrl+F |
  145. | new pane | ⌘N | Ctrl+N |
  146. | new window | ⌘⇧N | Ctrl+Shift+N |
  147. | agents | ⌘⇧A | Ctrl+Shift+A |
  148. | clear context | ⌘L | Ctrl+L |
  149. | compact context | ⌘M | Ctrl+M |
  150. | recent projects | ⌘R | Ctrl+R |
  151. | analytics | ⌘Y | Ctrl+Y |
  152. | split all panes | F8 | F8 |
  153. | combine panes | F7 | F7 |
  154. | vim mode | toggle in settings | toggle in settings |
  155. 50+ shortcuts total. every action accessible via keyboard on all platforms.
  156. ---
  157. ## agents
  158. yume includes 4 built-in agents synced to ~/.claude/agents/yume-*.md + 4 concurrent background agents:
  159. **foreground agents** (follows selected model):
  160. - **architect** - plans, decomposes, todowrite. runs before complex tasks.
  161. - **explorer** - read-only glob/grep/read. understands codebase.
  162. - **implementer** - edit/write. focused changes, minimal diffs.
  163. - **guardian** - reviews for bugs, security, performance.
  164. **background agents** (4 concurrent, async):
  165. - git branch isolation: yume-async-{type}-{id}
  166. - 30-minute timeout per agent
  167. - output to ~/.yume/agent-output/
  168. - streaming isolation from main session
  169. - merge conflict detection
  170. - worktree isolation: full filesystem isolation
  171. - permission modes + memory scope
  172. ---
  173. ## requirements
  174. - **claude account** - pro/max subscription or api key for claude code access (yume adds no markup)
  175. - **macos** 11+ (big sur or later) — arm64 + x64
  176. - **windows** 10+ (64-bit) — exe installer
  177. - **linux** ubuntu 20.04+ or equivalent — deb + rpm + flatpak
  178. ---
  179. ## license
  180. yume is free to download and use (demo: 2 tabs, 2 panes, 1 window). upgrade to pro for 99 tabs, 99 panes, 99 windows.
  181. - **monthly** — $4/mo, paypal subscription, cancel anytime, access until period ends
  182. - **lifetime** — $49 one-time payment, forever updates
  183. license is re-verified weekly with a 7-day network grace period. on validation failure, yume reverts to demo and offers a retry banner (key is retained).
  184. ---
  185. ## links
  186. - **download**: [aofp.github.io/yume](https://aofp.github.io/yume)
  187. - **docs**: [aofp.github.io/yume/docs/](https://aofp.github.io/yume/docs/)
  188. - **issues**: [github.com/aofp/yume/issues](https://github.com/aofp/yume/issues)
  189. - **discord**: [discord.gg/RhWVqCUnCc](https://discord.gg/RhWVqCUnCc)
  190. - **claude code cli**: [github.com/anthropics/claude-code](https://github.com/anthropics/claude-code)
  191. ---
  192. <div align="center">
  193. <p>made with claude code + yume</p>
  194. </div>

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.