Brilliant strangers
Agents don't know what changed since the last conversation. Every session starts with re-explaining a world that mostly did not move.
One MCP server: connect any agent to your or your users' context lake.
MCP is the interface. Your context lake is the product.
Fulcra is the user-owned context backend for AI agents. What accumulates for you is a context lake: real-world data, recorded knowledge, files, and agent work — and any AI that speaks MCP can read from it, write back to it, and coordinate with your other agents through your personal MCP server.
Point any MCP client at the URL below and sign in — no cables to plug in, no servers to babysit. The MCP server authenticates on its own, so a Life API token from the CLI will not open it.
MCP URL
Copy
https://mcp.fulcradynamics.com/mcp
You & Your Context
Your Agent
Fulcra Personal MCP
Fulcra Life API
Fulcra Datastore
Context App
Our MCP server solves your agents' context problem.
Modern life scatters context across silos, with four big challenges:
Nothing builds on anything
One agent can't build on what another agent learned. Each keeps its own private record of you, invisible to the rest.
Learning stays trapped
What an agent does learn is usually trapped inside one app, one agent, or one model provider.
And without a clear view of what an AI can read, people can't share confidently — or revoke cleanly.
Switch tools, start over
Change agents or models and the relationship resets. Every new AI product meets you as a stranger — unless the context is yours and travels with you.
TOOLS
Every loop starts with what changed.
The tools mirror the primitives — records and events, time series, files, shares — and the first one sets up the rest: ask what arrived since the agent last looked, including what your other agents wrote, then query only what changed.
What changed since the last loop
Ask what arrived in the account over any window — which kinds of records came in, and which files changed. It tracks when data reached Fulcra, not the records' own timestamps — last night's records land this morning — so an agent asking what's new since it last ran gets the right answer, including new work from your other agents, then queries only what changed.
Read the lake
Start from the catalog — it says what exists, grouped by the tool that reads it, and data other people have shared with you appears in that same catalog rather than somewhere separate. Then read records, time series, events, and files. Query the catalog, not a provider.
Write back
Record to any recordable type, store files that keep their native form, and define new custom types when the context calls for them. What an agent records lands where you — and every other agent you connect — can see it and build on it.
Private by design
Records and files are encrypted at rest, and nothing enters the lake unless you connect, record, or upload it. Every tool call runs as you, against your own account, over OAuth2 — agents are clients of the context, never its owners. Fulcra carries the security burden; you keep ownership and control.
PRIVACY & CONTROL
Your context, under your control.
Fulcra was designed by people who get privacy and know the importance of an infrastructure solution that can be the secure private datastore for the rest of your life. Sharing context with an AI should be a two-way door: grant access, revoke it, change agents, keep the context — your data is only shared with the people and tools you choose.
User-controlled sharing. OAuth2-protected access. Records, time series, event streams, and files from the sources you choose to connect.
.png){kind=logo}
