Commander Plus
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Commander Plus Register this project as a workspace and check its health"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Commander Plus
A small, local-first MCP control-plane core for AI agents that need persistent project context without sending workstation state to a hosted orchestration service.
This public repository contains the safe reference layer: workspace registration, reusable skill routing and health guardians. Production-only desktop automation, authenticated browser control, private relay infrastructure and local operational data are intentionally not included.
Why it exists
Most agent demos are stateless. Commander Plus treats the workstation as a durable execution environment: projects have identities, skills can be routed by intent, and health checks make local state explicit.
Related MCP server: BirdEye
Included
MCP server over stdio
Persistent local workspace registry
Reusable skill registry and deterministic intent routing
Workspace health guardian
JSON storage under a configurable local application directory
Node test coverage for routing behavior
Run
Install dependencies with npm install.
Build with npm run build.
Start with npm start.
Set COMMANDER_PLUS_HOME to move the local state directory.
Architecture
Client -> MCP server -> workspace / skill / guardian services -> local JSON state
The public core is intentionally conservative. It does not expose shell execution, private browser sessions, credentials, customer data or machine-specific paths.
Public vs private boundary
Public: local control-plane primitives, MCP interface, routing model, tests and documentation.
Private: production browser bridge, privileged desktop automation, private plugins, relays, operational logs, local databases and deployment configuration.
Security
Treat any workstation-control plugin as privileged software. Review tool scopes before enabling write or execution capabilities. Never commit local state, environment files or credentials.
Available Tools
5 toolsguardian_workspace_healthA
Check whether registered workspace roots remain readable.
Inputs: no named parameters.
skill_routeC
Rank locally saved skills against a task intent.
Inputs: query (required).
skill_saveC
Save a reusable skill with routing metadata.
Inputs: name (required), description (required), triggers, capabilities.
workspace_listB
List registered local workspaces.
Inputs: no named parameters.
workspace_registerC
Register a local project workspace.
Inputs: name (required), root (required), kind.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Workspace register vs list are clearly paired but distinct, and skill_save vs skill_route are clearly separated by action. The only slight ambiguity is between workspace_list and guardian_workspace_health, which both report on registered workspaces from different angles (enumeration vs readability).
All names are snake_case verb_noun style (workspace_register, skill_save, skill_route) which is predictable. The outlier is guardian_workspace_health, which uses a subsystem prefix rather than a verb, breaking the pattern for that one tool.
Five tools is a compact, well-scoped set for workspace registration plus local skill routing, with no filler. It sits slightly on the thin side, particularly on the skill side where a single save/route pair carries the whole capability.
There is save and route for skills but no skill list, update, or delete, and workspace_register has no corresponding unregister/remove counterpart. The surface covers the happy path but leaves obvious lifecycle gaps an agent would hit when managing or cleaning up entries.
Maintenance
Related MCP Connectors
Control plane for autonomous software labor. Agents claim objectives over MCP with audit trail.
Hosted MCP server and Agent Skills for specs, engineering threads, and project memory.
Persistent memory, knowledge graph, task management and more for AI agents via MCP.
Private-by-default, local-first memory/context/task orchestrator for MCP apps and agents.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceMCP server for task management, project knowledge, workspace trust, runner sandboxes, extension registry, and workflow prompts, enabling AI agents to manage tasks and collaborate locally.5,117 npmApache 2.0
- AlicenseNot gradedqualityCmaintenanceA local-first mission control for AI agent harnesses, providing a unified MCP gateway for shared memory, task queue, and encrypted secrets across multiple agents.2MIT
- AlicenseAqualityCmaintenanceMCP server that enables running and managing file-based AI Skills, Agents, and Flow workflows as DAGs, with tools for listing, executing, and resuming tasks via stdio JSON-RPC.9MIT
- AlicenseNot gradedqualityBmaintenanceProvides a local-first AI workspace for chatting, invoking skills, managing projects, and integrating MCP tools, with all data stored on the user's machine.2Apache 2.0