ClearPath is a graph workspace that any AI can connect to via MCP. Your context, skills, tasks, and decisions live in the graph — persistent, structured, and reachable from any agent, any environment, any session.
Built for people doing serious work with AI. Also works great without it.
A graph where you and your agents share the same nodes, edges, and context. Not a chat window. Not a list. A persistent map of everything you're working on.
Agents pick up exactly where they left off. Context, decisions, and task state survive across sessions, environments, and agent instances.
Protocols and skills live as graph nodes — the same graph humans navigate daily. No separate prompt files. No drift. One source of truth.
Every node carries cost, duration, priority, status, due dates, and a full markdown editor for notes. Edges carry quantities. Derive reports and projections from the graph itself.
Any MCP-compatible agent can read, write, and navigate your graph. Claude, GPT, custom agents — they all connect to the same workspace.
Throw anything into The Net — ideas, tasks, links, decisions. It lands in a temporal column, ready for review. Agents can capture here too, with full provenance.
I built ClearPath for myself — to help with my own memory, systems design, and planning. No AI agents in mind at all.
Then I learned about MCP, and something clicked. Suddenly I had a way for me AND my agents to stay aligned through a shared surface: planning, reference, capture, pattern recognition, skill building.
The graph was already the right shape for agent memory. Reality confirmed the design — it wasn't planned that way. The tool became something fundamentally different without the underlying structure changing.
This isn't a pivot. It's a discovery.
Free while in early access. Join the waitlist to get started.