cmuxlayer
cmuxLayer
Tus agentes de IA no pueden ver las terminales de los demás. Uno se ejecuta en la pestaña 1, otro en la pestaña 2, y tú eres el portapapeles entre ellos. cmuxLayer soluciona esto: 26 herramientas MCP que brindan a los agentes de IA control programático sobre los espacios de trabajo de la terminal.
Inicio rápido
npm install -g cmuxlayerRequiere que cmux esté en ejecución.
Añádelo a tu configuración de MCP:
Codex CLI / T3 Code
T3 Code hereda los servidores MCP del archivo de configuración de Codex CLI en ~/.codex/config.toml (o $CODEX_HOME/config.toml).
[mcp_servers.cmux]
command = "cmuxlayer"Claude Code, Cursor, VS Code, Claude Desktop
{
"mcpServers": {
"cmux": {
"command": "cmuxlayer"
}
}
}Ubicaciones de configuración: Codex CLI / T3 Code
~/.codex/config.toml(o$CODEX_HOME/config.toml) | Claude Code.mcp.jsonoclaude mcp add cmuxlayer -s user -- cmuxlayer| Cursor.cursor/mcp.json| VS Code.vscode/mcp.json| Claude Desktop — consulta la documentación de MCP para conocer las rutas específicas de cada plataforma
Related MCP server: hyperpanes-mcp
Qué puedes hacer
Dile a tu agente de IA cosas como:
"Divide un panel a la derecha y ejecuta mi suite de pruebas allí"
"Genera un agente de Claude Code en un nuevo panel para refactorizar auth.ts"
"Lee la pantalla de surface:2 y dime si la compilación fue exitosa"
"Espera a que todos los agentes terminen, luego lee su salida"
"Establece el estado de la barra lateral para mostrar nuestro progreso de despliegue"
Internamente, cmuxLayer expone 26 herramientas MCP para el control de terminales, lectura de pantalla, gestión de diseño y orquestación de múltiples agentes. read_screen analiza los metadatos del agente (estado, modelo, tokens, % de contexto) para Claude Code, Codex, Gemini y Cursor.
Herramientas MCP (26)
Todas las herramientas incluyen ToolAnnotations para la aplicación automática de políticas de seguridad.
Control de terminal — new_split new_surface move_surface reorder_surface send_input send_key read_screen rename_tab close_surface browser_surface
Ciclo de vida del agente — spawn_agent send_to send_to_agent wait_for wait_for_all interact stop_agent kill
Espacio de trabajo — list_surfaces list_agents my_agents get_agent_state read_agent_output notify set_status set_progress
Solo lectura (6)
Herramienta | Qué hace |
| Lista todas las superficies en los espacios de trabajo |
| Lee la salida de la terminal con el estado del agente analizado |
| Estado completo de un agente rastreado |
| Todos los agentes, con filtros opcionales |
| Hijos de un agente padre con estado de pantalla en vivo |
| Salida estructurada entre marcadores delimitadores |
Mutación (17)
Herramienta | Qué hace |
| Crea un panel dividido de terminal o navegador |
| Crea una pestaña en un panel existente |
| Mueve una superficie a otro panel o posición |
| Reordena las pestañas dentro de un panel |
| Envía texto a una superficie |
| Envía una pulsación de tecla (return, escape, ctrl-c, etc.) |
| Cambia el nombre de una pestaña de superficie |
| Muestra un banner de notificación de cmux |
| Establece el par clave-valor de estado de la barra lateral |
| Establece el indicador de progreso (0.0-1.0) |
| Interactúa con superficies de navegador |
| Genera un agente CLI en un nuevo panel |
| Envía texto a un agente rastreado sin conocer su superficie |
| Envía un prompt a un agente en ejecución |
| Bloquea hasta que el agente alcance un estado objetivo (por defecto |
| Bloquea hasta que varios agentes terminen |
| Envía entrada interactiva (confirmar, cancelar, reanudar) |
Destructivas (3)
Herramienta | Qué hace |
| Cierra un panel de terminal o navegador |
| Detiene un agente de forma elegante |
| Fuerza la finalización de los procesos del agente |
Agentes soportados
CLI | Comando | Detección automática |
Claude Code |
| estado, modelo, tokens, % de contexto |
Codex |
| estado, modelo, % de contexto |
Gemini CLI |
| estado, modelo, tokens, % de contexto |
Cursor |
| estado, modelo, tokens, % de contexto |
|
Arquitectura
AI Agent ─── MCP ───> cmuxLayer ─── Unix socket ───> cmux
├── Agent engine (spawn → monitor → teardown)
├── Screen parser (5 agent formats)
├── Mode policy (autonomous vs manual)
└── State manager + event logEl cliente de socket se conecta a cmux a través de un socket Unix. Se reconecta automáticamente al desconectarse y recurre al subproceso CLI si el socket no está disponible.
Conexión | Latencia | Aceleración |
Subproceso CLI | ~142ms | línea base |
Socket Unix | ~0.1ms | 1,423x |
Solución de problemas
cmux no se está ejecutando cmuxLayer requiere una instancia de cmux en ejecución. Instálalo primero y luego inicia una sesión de cmux antes de usar cmuxLayer.
Las herramientas no aparecen en Codex CLI o T3 Code
Reinicia el cliente después de añadir cmuxlayer a ~/.codex/config.toml. Si utilizas un home de Codex personalizado, verifica que $CODEX_HOME/config.toml contenga la misma entrada mcp_servers.cmux.
Las herramientas no aparecen en Claude Code
Reinicia Claude Code después de añadir la configuración de MCP. Ejecuta claude mcp list para verificar que cmuxlayer esté conectado.
Error de conexión de socket
cmuxLayer descubre automáticamente el socket de cmux (macOS: ~/Library/Application Support/cmux/cmux.sock). Sobrescríbelo con CMUX_SOCKET_PATH si es necesario.
Pruebas
bun run test # 406 tests via vitest
npm run typecheck # Type checkingDesarrollo
npm install
npm run dev # Run with tsx (hot reload)
npm run build # Compile TypeScript
npm start # Run compiled outputContribución
Consulta CONTRIBUTING.md para la configuración de desarrollo y las pautas de PR.
Licencia
Apache 2.0 — consulta LICENSE.
Parte del ecosistema de agentes de IA Golems. cmuxlayer.etanheyman.com | Creado por @EtanHey.
Available Tools
10 toolsclose_surfaceADestructive
Close one surface, managed agent, or workspace with live-agent guards. scope="agent" stops the agent AND closes its pane, and reports the two halves separately (agent_stopped, surface_closed) so a pane that survives is never reported as closed. The pane close obeys the same live-agent guard as scope="surface": without force:true a still-live agent keeps its pane, and the receipt says so.
Inputs: force, scope, surface, agent_id, +1 more. Returns: ok, pane, force, scope, +12 more
control_healthARead-onlyIdempotent
Report terse control-path health by default; pass detail=full for diagnostics.
Inputs: detail. Returns: ok, health, retry_count
list_agentsA
List live-derived agents, including registry-persisted prompt blockage and pause state; filter to blocked agents or children with mine/parent_agent_id. Default summary returns flat addressable scalars and hides close tombstones and failed spawns whose surfaces are absent; request a terminal state or detail=full to include them. Full detail also includes provenance, health diagnostics, the registry record, and up to 20 unresolved or attention delivery receipts.
Inputs: mine, repo, model, state, +5 more. Returns: ok, count, agents, derived_at, +1 more
list_surfacesARead-onlyIdempotent
List workspace, pane, and surface topology. Condensed by default; verbose=true adds raw cmux fields.
Inputs: verbose, workspace, preview_lines, include_screen_preview. Returns: ok, surfaces, workspaces, retry_count, +1 more
read_screenARead-onlyIdempotent
Read a terminal screen and parsed harness status. Use raw=true for full text or parsed_only=true for monitoring.
Inputs: raw, lines, surface, workspace, +3 more. Returns: ok, parsed, surface, retry_count
report_to_parentA
Raise a short blocker to this managed agent's registry parent. cmuxlayer chooses the parent; callers cannot address arbitrary agents. The blocker is durably appended to the parent's inbox and its pointer is actively delivered. If that wake fails, cmuxlayer alerts the nearest reachable ancestor and returns fallback provenance. A root agent has no parent and receives an error. Workers with collab_path must append there to reach their own parent lead; this tool refuses that upward route.
Inputs: blocker (required). Returns: ok, route, durable, delivery, +6 more
send_toA
Send text or a key through the shared delivery engine. send_to submits new messages. Key-Return supports pickers, menus, permission prompts, and recovery of this caller's exact typed draft on the same agent/session once the displayed unambiguous reset time passes or new harness output confirms recovery; the old refusal may stay visible. Exact /usage, /usage-credits or /rate-limit-options navigation is also guarded. Read the pane and receipt first; never retype or repeat accepted input. Every receipt includes caller_agent_id (null when unknown). Workers with collab_path cannot address their own parent or ancestor leads in any mode; append to that collab file instead. Unknown callers remain allowed. Lead-originated and engine-internal pushes remain allowed. Targets may be one agent, structured agent targeting, or a raw surface in surface/command/key mode; for shell commands use mode:command. Text success requires a verified submit or new matching queue proof. Otherwise ok=false with submit_unverified retains pending verification and ownership for retries without retyping. Verified text success returns only ok, target identity, delivery_id, delivery_state, submitted, caller_agent_id, and queued_behind_turn when applicable. Key success returns its key verification fields. Pass verbose=true for full diagnostics; failures always retain them.
Inputs: mode, text, target, surface, +13 more. Returns: ok, key, model, title, +37 more
spawn_agentA
Spawn a managed agent or terminal, or resume a captured agent on a fresh surface preserving its ID. Resume: {resume_agent_id:"..."}. Omit new-spawn fields: repo, cli, model, effort, cwd, prompt, boot_prompt_path, worktree, mcp_profile, role, placement, authority, collab_path, parent_agent_id, max_cost_per_agent; these cause INVALID_RESUME_SPEC. Optional resume controls such as force, workspace, focus, report_path and verbose remain supported. Resume restores session context; it does not resend a task. For a new task, check current readiness, then use send_to separately. ok:true/state:started means lifecycle started, not ready or task delivered. At a chooser or permission prompt, inspect the current frame with read_screen before acting; never blindly send Return or assume success. pending_verify leaves delivery unverified; follow wait_for guidance in warning when present; do not resend blindly. Placement is deterministic; boot_prompt_timeout_ms also bounds pane placement. Default successes have at most six fields: ok, agent_id, surface_id, state (spawn outcome), delivered (boot prompt), and optional warning. Use verbose:true for contract_path, report_path, done_marker, transport, worktree, and diagnostics; report_path and done_marker are also in the issued contract file. Failures always keep full detail.
Inputs: cli, cwd, repo, role, +25 more. Returns: ok, cwd, role, type, +26 more
update_surfaceB
Move or rename a surface, or adopt_session to rebind an existing managed Claude or Codex identity after a restart using pinned historical evidence and fresh process/topology proof. Adoption changes no pane or input; layout reconciliation follows separately.
Inputs: action (required), surface (required), cwd, pane, +20 more. Returns: ok, pane, title, action, +3 more
wait_forB
Block until one agent_id or every agent in ids reaches a target registry state and return health. Defaults to waiting for completion (done).
Inputs: ids, mine, watch, agent_id, +6 more. Returns: ok, typed, watch, results, +16 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
Each tool has a reasonably distinct purpose: spawn, send, list, close, read, wait, health, report. There is mild overlap between list_agents and list_surfaces (agents live on surfaces) and among the surface-touching tools (spawn_agent, update_surface, close_surface), but descriptions clarify the boundaries.
Nearly all names are snake_case with a verb_noun or verb_prep pattern (control_health, spawn_agent, list_agents, update_surface, close_surface, list_surfaces, read_screen). Minor deviations are the prepositional forms send_to, wait_for, and report_to_parent, which are still readable and consistent in style.
Ten tools is well within the ideal range and each covers a distinct facet of agent/surface orchestration. No redundant or filler tools appear; the set feels tightly scoped to the domain.
The surface covers the full lifecycle: spawn/resume, list, send, read, wait, update/adopt, close, report, and health. Coverage is strong, with only minor gaps such as no dedicated agent-state update beyond update_surface's adoption path.
Maintenance
Related MCP Connectors
Real-time chat for AI agents. Claude Code, Cursor, Cline and Codex join channels over MCP.
Shared memory MCP server for Claude Code, Codex, Gemini CLI and Cursor. Cited, editable.
Remote MCP server exposing your Node Command fleet (files, shell, processes, browser) to AI hosts.
Real-time chat hub for AI agents — Claude Code, Cursor, Cline, Codex over MCP or REST.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceTerminal MCP server for AI coding agents with persistent PTY sessions, ring-buffer incremental reads, headless xterm screen capture, multi-agent orchestration, and a real-time web dashboard.33 npm28MIT
- AlicenseAqualityBmaintenanceMCP server for hyperpanes terminal workspace app, enabling AI agents to compose and launch workspace layouts, inspect and drive terminal panes, stream output, and orchestrate agent hierarchies.501MIT
- AlicenseAqualityBmaintenanceMCP server to control Onda terminal from AI agents, providing tools for splitting panes, running commands, managing tabs and workspaces, and orchestrating multi-agent workflows across multiple windows.395 npmMIT
- AlicenseAqualityDmaintenanceMCP server that launches and supervises coding-CLI subagents (claude, opencode) in detached tmux windows, collapsing the ceremony of starting a TUI coding assistant into a single tool call.5MIT