KeystoneMCP is a local-first production-context and project-management MCP server that gives AI assistants persistent memory for creative projects.
Connect it to Claude, Codex, or another MCP-compatible AI client to provide reliable context about projects, tasks, assets, versions, dependencies, workflow rules, approvals, and next production steps.
KeystoneMCP can create profile-based project folder structures, establish naming and path rules, and tell the AI exactly where source files, working files, reviews, publications, and deliverables should be saved.
It acts as the authoritative project-context layer across sessions. It does not replace your creative software or aggregate every tool. Your AI consults KeystoneMCP for production state while connecting directly to Houdini, Maya, Nuke, Blender, ComfyUI, and other tool MCPs.
Suitable for VFX, animation, game development, graphic design, photography, marketing, and other creative-production workflows.
What It Does
KeystoneMCP helps an AI assistant understand:
- What projects, assets, shots, deliverables, and tasks exist
- Which versions are current or approved
- What depends on what
- Which work is blocked or ready
- How project folders should be structured
- Where source, work, review, publication, and delivery files should be saved
- What the next valid production step is could be affected when something changes
Your AI client consults KeystoneMCP for project context, uses creative-tool MCPs directly, and records completed work back into persistent production state.
Try it now
The launch version is free for eligible indie users to claim and keep for the next month.
Iād especially appreciate feedback on where project organization creates the most friction in your own AI-assisted workflows. Thanks for taking a look!
Top comments (0)