Personal context. In your keeping.

Switch agents.
Keep the thread.

Your preferences, projects, and decisions shouldn’t start over with every new assistant.

Threadkeeper gives the agents you use one shared memory. Yours to inspect, correct, and carry forward.

Explore on GitHub Read the pitch

Early preview · Self-hosted · Built for MCP

THE CONTINUITY ENGINENo. 001

◆ ONE MEMORY, ACROSS YOUR AGENTS

SAVED BY YOUR WRITING ASSISTANT01 / 01

Preference · Writing

“Keep it clear, concise,
and in my own voice.”

◇ You said thisAvailable to your agents

A preference saved once. A thread that continues.

ILLUSTRATIVE DEMOSELECT AN AGENT ABOVE

Remember what mattersKnow where it came fromKeep the final say

01 / THE IDEA

A little less
starting over.

A new assistant shouldn’t mean explaining yourself all over again. Let your context travel with you.

01

Tell it once.

An assistant saves a lasting preference, a project decision, or a detail you want remembered.

Connect an assistant

02

Keep the context.

Threadkeeper keeps the memory with its original source. You can see exactly what it holds, and why.

Explore the Record

03

Pick up the thread.

The next connected agent can recall relevant context within its permissions, without seeing the other agent’s conversation.

See your controls

02 / THE RECORD

Memory you can
actually see.

A personal archive, with you at the controls. Browse what’s remembered, inspect its source, and change your mind.

Every memory, its source, and your corrections. One place to keep it all straight.

ACTUAL PRODUCT UI · EXAMPLE DATA

03 / IN YOUR KEEPING

An assistant can learn.
You get the final word.

Personal memory should be understandable, editable, and yours to take with you.

A source for every memory.

See the original words. Distinguish what you said from what an agent reported or inferred.

Your corrections take effect.

Fix a detail and the next recall uses your version. The previous interpretation stays in history.

Access on your terms.

Give each assistant its own credential, choose project access, and decide whether it can read or save.

Yours to keep. Yours to forget.

Export your memory to another installation, or remove it from future recall. Copies already shared outside Threadkeeper remain outside its control.

WHY THREADKEEPER EXISTS

“What if the context that makes one agent feel personal could move with you to the next?”

It started with how well ChatGPT remembered personal details and preferences. Threadkeeper brings that idea to a memory layer you can use across agents, without the long introduction every time.

◆ Drake Thomsen Creator, Threadkeeper

SAME YOU. NEXT AGENT.

Your context has a history.
Give it a home.

Explore the code, run the local demo,
and help shape what comes next.

Explore on GitHub Run the demo

Early preview. Runs on your computer or server.
No hosted service yet. Open-source license to be determined.

Switch agents.Keep the thread.

A little lessstarting over.