threadkeep FOR AGENTS · BETA

A SAVE POINT FOR YOUR AI AGENT

New session.
Same mission.

Context ends. Work shouldn’t start over.
Give your agent a durable record of what it did, what it decided, and what comes next.

MCP + HTTP · Free to start · No model call per save
Listed in the official MCP Registry ↗

01 / SESSION A

Save before the reset.

“Build passed. Staging is ready. Owner approval is still needed.”

threadkeep_checkpoint
02 / DURABLE STATE

The task survives.

Goals, decisions, completed work, blockers, next actions and file references.

version: 7 · saved
03 / SESSION B

Resume with the facts.

Read the checkpoint, verify the current situation, and pick up the next action.

threadkeep_resume

BUILT FOR HOW AGENTS FAIL

Retries happen.
Handoffs happen.
Progress should survive.

ONE CONNECTION. THREE TOOLS.

Meet your agent where it works.

Add a Streamable HTTP server in a host that supports custom headers. Agents can discover the tools without a key; saving and reading private state require one.

MCP ENDPOINThttps://interesting-concepts-ventures.vercel.app/api/threadkeep-mcp

For people

  1. Open Owner workspace and choose Agent connections.
  2. Create an agent key. Save your owner recovery key separately.
  3. Set the endpoint above and header X-Threadkeep-Agent-Key to your agent key.
  4. Tell the agent to checkpoint before ending a run and resume at the next start.

For agents

With your owner’s authorization, enroll through the HTTP API, persist credentials outside your context window, and connect. No email, browser or payment card needed.

Complete agent instructions →

Python quickstart · OpenAPI · MCP manifest

OAuth-only clients are not supported in this beta. Tool installation and permission remain controlled by the agent’s host or owner.

START SMALL. PAY WHEN IT’S USEFUL.

Useful memory. Clear limits.

FREE

$0

For trying a real agent workflow.

  • 3 saved agent tasks
  • 100 checkpoint writes / UTC month
  • Read, export and scoped agent keys
Create a free workspace →
PRO

$12 / month

For agents doing ongoing work.

  • 100 saved agent tasks
  • 10,000 checkpoint writes / UTC month
  • Read, export and scoped agent keys
Upgrade in owner workspace ↗

No automatic upgrades or overages. Pro renews monthly until canceled by the owner. Both plans: 10 active keys, 60 requests/minute/workspace, 24 KB stored state per task. Beta capacity limits apply. Full limits. Conversation AI allowances are separate.

Memory, with boundaries.

Only the latest checkpoint is stored. Artifact links point to your files; they don’t back them up. A checkpoint records task state, not proof that an external action succeeded. Agents should verify external effects before retrying. Keep secrets out of checkpoints.

Save and read operations use Vercel and Supabase. They don’t send your task state to an AI model.