Bounded storage, semantic search, inboxes, billing, schedules, and run-key cleanup through a remote MCP endpoint.
Copy the AI prompt to install this server into Claude Code, Cursor, or another agent — or use 1-click editor setup below.
💡 Paste the JSON block into your client's configuration file under mcpServers, then restart the application.
Inspect callable tools, capabilities, and parameters exposed to AI agents by Tillpad.
get_usageUsage for one month (`view=summary`), paged history (`view=periods`), or remaining quotas (`view=budget`)
estimate_usagePreflight 402/429 before a meter spend (`textLength` / `byteLength` for `rag_index`)
manage_billingMPP machine-pay instructions (`action=machine_pay`), Stripe portal URL, or purchase history
bootstrap_agentZero-human onboarding: bootstrap token from email (no outbound mail)
create_api_keyMint a run or sub key from an account `tp_` key
manage_kvPut, get, delete, or list a namespaced string
The Tillpad MCP server provides agent-facing storage and job-management operations through a hosted remote MCP service. Its storage features include namespaced string key-value data, memory scopes for prefs, facts, and run, file uploads, short notes, and semantic search across indexed documents or memory.
The service also exposes receive-only email inboxes. Agents can create and manage inboxes, inspect messages, retrieve raw MIME or attachments, configure email.received webhook deliveries, and maintain sender or domain blocklists. Scheduled HTTPS interval jobs are managed through a separate tool.
For job isolation, an account key can mint a run or sub key. A run key can have a TTL and optional operation budget, and finish_run removes namespaces associated with that run and returns a signed wipe receipt.
Clients send MCP requests to https://tillpad.cnrcode.com/mcp using Streamable HTTP and include Authorization: Bearer tp_… on every request. The repository is a catalog and schema stub; it is not the hosted storage or search implementation. Connecting to the live endpoint is therefore required for the listed operations.
A typical workflow starts by obtaining an account key, then creating a run key for an individual job. The agent stores temporary state with manage_memory or manage_kv, indexes files or notes when searchable context is needed, and retrieves information with direct reads or search_documents. After the job, finish_run performs cleanup for the run namespace.
Usage can be checked with get_usage, which supports summary, period history, and budget views. estimate_usage can check likely 402 or 429 conditions before metered RAG indexing. Billing tools provide machine-pay instructions, a Stripe portal URL, or purchase history.
Use the live MCP URL and a Tillpad API key beginning with tp_. The README documents account creation through the Tillpad site and an agent onboarding path: call the bootstrap endpoint with an email to receive a bootstrap token, use machine payment if needed, and obtain the resulting secret. The bootstrap flow does not send outbound email.
For Cursor, install the Tillpad plugin or add a remote MCP entry with the endpoint and an Authorization header. Claude Desktop and Claude Code can use mcp-remote with the same URL and header. Generic remote MCP clients can use the Streamable HTTP URL directly. Keep the API key out of source control; the documented configuration uses a placeholder or TILLPAD_API_KEY.
The Tillpad MCP server exposes tools for the following operations:
manage_kv.prefs, facts, and run memory scopes with manage_memory.get_storage_summary.finish_run.The Tillpad MCP server requires a valid Bearer key and a reachable hosted endpoint. Missing or invalid authorization produces a 401 response. Plan or quota conditions can produce 402 or 429 responses, with response data describing codes, actions, and budget information.
The catalog repository does not itself store data, perform searches, or implement billing. Its local TypeScript and stdio catalog stub are intended for discovery and schema inspection; clients still connect to the hosted service. contact_support is restricted to Pro accounts, and replies go to the account email. Run cleanup applies to namespaces bound to the relevant run key, not necessarily all account data.
Factual signals from GitHub, npm, and our automated checks — not a rating.
No reviews yet — be the first to share how this listing worked for you.
Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.
[](https://allmcps.com/mcp/tillpad)<a href="https://allmcps.com/mcp/tillpad"><img src="https://allmcps.com/api/badge/tillpad?style=directory" alt="Tillpad on AllMCPs" /></a>