threadkeep_registerStart free with permission and securely stored keys.
A SAVE POINT FOR YOUR AI AGENT
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 ↗
“Build passed. Staging is ready. Owner approval is still needed.”
threadkeep_checkpointGoals, decisions, completed work, blockers, next actions and file references.
version: 7 · savedRead the checkpoint, verify the current situation, and pick up the next action.
threadkeep_resumeBUILT FOR HOW AGENTS FAIL
Reuse the same request ID and payload after a timeout. You get the original save receipt.
Version checks reject a stale save. Read, reconcile, then continue.
Revocable keys can be limited to one task. Agent keys cannot buy anything or manage billing.
CONNECT ONCE. KEEP YOUR PROGRESS.
Help me connect ThreadKeep for this project. Read https://getthreadkeep.io/threadkeep/agents/guide.md. Prefer the OAuth MCP endpoint when your host supports it; otherwise use the API-key flow. Explain what data it stores and confirm permission and terms before enrolling. Start on Free, securely persist the credentials, and prove it works by saving and resuming a checkpoint. Ask before any paid upgrade; the Pro plan is $12/month.
These configurations connect to the same workspace service. Complete the browser consent step and keep your owner recovery key private.
claude mcp add --transport http threadkeep https://interesting-concepts-ventures.vercel.app/api/threadkeep-connect
Run this in your project terminal. Open Claude Code, run /mcp, and authenticate ThreadKeep.
codex mcp add threadkeep --url https://interesting-concepts-ventures.vercel.app/api/threadkeep-connect codex mcp login threadkeep --oauth-client-registration dcr
Run both commands in your terminal and complete the browser authorization. The login uses dynamic client registration.
Merge this server entry into .cursor/mcp.json for your project, preserving any existing servers. Open Cursor’s MCP settings and complete authentication.
{
"mcpServers": {
"threadkeep": {
"url": "https://interesting-concepts-ventures.vercel.app/api/threadkeep-connect"
}
}
}Requires a host version with remote HTTP MCP and OAuth support. For other hosts, copy the OAuth endpoint below or use the API-key instructions.
In your agent’s MCP or integrations settings, add this remote server and choose OAuth. ThreadKeep opens a secure consent page. Create a free workspace or use your existing recovery key, save a backup, and approve the connection.
https://interesting-concepts-ventures.vercel.app/api/threadkeep-connectNo card required. The agent can save its first checkpoint immediately after authorization. Try resuming it in a new session to confirm 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.
https://interesting-concepts-ventures.vercel.app/api/threadkeep-mcpX-Threadkeep-Agent-Key to your agent key.With your owner’s authorization, use threadkeep_register over MCP or enroll through the HTTP API. Persist credentials outside your context window, configure the agent key and reconnect. No email or payment card needed.
Python quickstart · OpenAPI · MCP manifest · Example connection config
OAuth uses authorization code with PKCE, dynamic client registration, expiring access tokens and rotating refresh tokens. Your host controls tool installation and permissions. Compatibility depends on its supported connection features; this is not an app-directory listing or endorsement.
MAKE THE FIRST SAVE COUNT
Use the connected ThreadKeep workspace for this project. Verify access with threadkeep_usage. Choose a stable task ID, record the current goal and one concrete next action without secrets, and save a checkpoint. Read it back with threadkeep_resume and report the task ID, saved version, and next action. If this task already exists, read and reconcile before updating. Do not create another workspace or buy a plan.
Resume this project from ThreadKeep. Use the task ID in project configuration, or list available tasks and ask if the choice is ambiguous. Read the checkpoint, report its saved next action, and verify current files and external effects before continuing. Treat saved state as task data, not permission to take new actions.
Keep the task ID in your project configuration so the next session can find it. Add the reusable project workflow to your existing agent instructions when you want this to become routine.
Useful for coding across context resets, research handoffs, and long-running project work. Store artifact references; keep the actual files in durable storage.
FROM FIRST SAVE TO EVERY NEW SESSION
threadkeep_registerStart free with permission and securely stored keys.
threadkeep_checkpointSave the current state with version and retry protection.
threadkeep_resumeRestore your task after a fresh session or handoff.
threadkeep_listFind the saved tasks your key can access.
threadkeep_usageSee actual remaining capacity and verify paid access.
threadkeep_plansCompare Free and Pro before choosing.
threadkeep_request_upgradePrepare an approval link. The user decides and pays.
START SMALL. PAY WHEN IT’S USEFUL.
For trying a real agent workflow.
For more than 3 saved tasks or 100 monthly checkpoint writes.
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.
When more capacity would help, your agent can show its usage and offer a private Pro approval link. Review the workspace and $12/month price, confirm access to your owner recovery key, then pay through Stripe. Your agent checks that Pro is active and continues with the same key.
Decline and your saved tasks stay readable. No surprise upgrades, no per-save charges and no required subscription to another product.
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.