Persistent production state
Keeps projects, tasks, versions, dependencies, workflow state, and audit history available across sessions so the AI can continue from the same production context.
KeystoneMCP
KeystoneMCP is a local-first production-context MCP server that helps AI assistants track creative projects, tasks, versions, dependencies, and workflow rules across sessions. It is aimed at creative production workflows that need persistent project state alongside separate tool-specific MCP connections.
Overview
KeystoneMCP is a local-first production-context and project-management MCP server for creative workflows. It gives AI assistants persistent access to project state so they can reason about projects, assets, tasks, versions, dependencies, approvals, and next production steps without relying on memory from a single chat session.
The product is meant to act as the authoritative project-context layer while your AI client connects separately to creative tools such as Houdini, Maya, Nuke, Blender, and ComfyUI through their own MCP servers. It supports profile-based project folder creation, existing-project folder import, and filesystem-first storage on local drives, external drives, or mounted NAS locations.
Key features
Structured production graph
Represents assets, production units, tasks, versions, and dependencies as a structured graph, which helps the AI understand what depends on what and what is blocked or ready.
Version and workflow tracking
Tracks current and approved versions, preflight checks, blockers, and downstream impact so the AI can suggest the next valid step instead of guessing.
Folder and file role rules
Defines naming, path, and folder rules for source, work, review, publication, and delivery files, with profile-based project and folder creation.
Local production console
Provides a browser-based console with search, filtering, pagination, and light, dark, or OS themes for working with larger projects locally.
Local and trusted remote access
Supports local stdio MCP plus authenticated remote MCP for trusted local networks, with named viewer, editor, and administrator accounts and personal, revocable tokens.
Common use cases
Long-running creative projects
Keep a creative project’s assets, shots, tasks, and version state available to an AI assistant across multiple sessions, so follow-up work can use the same project context.
Project setup and file organization
Set up project folder structures and file-role rules for source, working, review, publication, and delivery files before production starts, reducing ambiguity about where files belong.
Preflight and change impact checks
Check blockers, dependencies, current versions, and downstream impact before making changes so the AI can recommend the next valid production step.
Managing larger local projects
Use the browser console to inspect larger projects, search for entries, and filter or paginate results when the production graph grows beyond a simple list.
Multi-tool creative pipelines
Connect an MCP-compatible AI client to KeystoneMCP while using separate MCP tools for Houdini, Maya, Nuke, Blender, or ComfyUI, with KeystoneMCP providing the shared production context.
Pros and Cons
Pros
- Keeps production metadata local instead of using a hosted data-collection service.
- Works across sessions, which is useful when AI assistance needs durable project context.
- Supports both creative planning and project organization through graphs, workflow templates, blockers, and path rules.
- Includes a web UI for browsing, searching, filtering, and managing larger projects.
- Offers separate support for local stdio MCP and authenticated remote MCP access for trusted networks.
Cons
- It is not a full replacement for creative software or third-party tool MCP servers; it only provides the production-context layer.
- Windows users currently require WSL.
- Single-seat licences do not include a standard trial; evaluation is handled through purchase and possible refund, or by requesting an evaluation copy for multi-seat customers.
- Remote access should only be enabled on trusted private networks, and public internet exposure is not recommended.
FAQ
Which AI clients can use KeystoneMCP?
It is designed to work with any MCP-compatible AI client, including Claude and Codex, as long as the client can connect to a local or authenticated remote MCP server.
Can KeystoneMCP be used over a network?
The source describes local stdio MCP access and authenticated remote MCP access. Remote and LAN modes are intended for trusted private networks, and LAN transport uses HTTP by default.
Where is project data stored?
The source says KeystoneMCP stores production metadata locally in your configured project folders and keeps LAN authentication information in its local configuration directory.
Is there a trial or evaluation option?
No separate trial is offered for single-seat licences. Single-seat users are directed to purchase the Indie plan first, while studios and other multi-seat customers may request an evaluation copy by contacting the vendor.
Can I modify or redistribute the software?
KeystoneMCP is distributed in source-readable form because the Python service and Web UI are installed locally. The licence allows reasonable modification for the licensee’s own internal use, but not redistribution or resale.
Quick Facts
- Category
- Production-context MCP server
- Platform
- Local Python service and web UI
- Primary use
- Creative production project management for AI assistants
- Supported environments
- macOS, Linux, and Windows via WSL
- AI compatibility
- Any MCP-compatible client
- Storage model
- Filesystem-first, stored locally
Alternative a KeystoneMCP
Struere
Struere is an AI-native platform for turning spreadsheet data into structured operational software with dashboards, alerts, and automations. It is aimed at teams that want to replace manual spreadsheet workflows without building custom tools from scratch.
runevr
runevr is a connected workspace for creative teams that combines boards, media review, chat, recordings, drive, calls, and AI agents in one room-based environment. It is designed to reduce tool switching and keep project context together.
Goldfish
Goldfish is an AI memory app for macOS that helps you reply, write, summarize, and continue using the context already on your Mac. It stores memory locally and writes in your tone.
Lasso
Lasso is an ecommerce product data platform for enriching catalog records, processing supplier files, generating product content, and monitoring competitors. It combines a web app with a REST API, SDK, and MCP server for teams and developers.
Biji
Biji è una piattaforma versatile progettata per migliorare la produttività attraverso strumenti e funzionalità innovative.
Codex Plugins
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.