# spendctl

> spendctl is the spend-policy control plane for AI agents: budgets, approvals, receipts and chain tracing across every wallet and every HTTP 402 protocol — enforced in a sidecar you run, never by holding your keys or funds.

Important notes:

- Agents pay for APIs over HTTP 402 (x402 and MPP). spendctl sits between the agent and the seller as a local sidecar: it enforces a budget per agent, allowlisted hosts and pinned recipients, holds anything above a threshold for a human approval, and writes a receipt for every settlement.
- The fastest way to see it is the hosted sandbox MCP server at https://mcp.spendctl.dev/mcp (mock money, no login). The local sidecar runs with `npx spendctl init && npx spendctl start` and is free for two agents; Pro and Team add the hosted control plane.
- The documentation below is the English source; a Japanese edition lives under https://spendctl.dev/ja/docs/.

## Documentation Sets

- [Abridged documentation](https://spendctl.dev/llms-small.txt): a compact version of the documentation for spendctl, with non-essential content removed
- [Complete documentation](https://spendctl.dev/llms-full.txt): the full documentation for spendctl

## Notes

- The complete documentation includes all content from the official documentation
- The content is automatically generated from the same source as the official documentation

## Optional

- [Landing page](https://spendctl.dev/): product overview, pricing and FAQ
- [Hosted sandbox MCP server](https://mcp.spendctl.dev/mcp): a remote MCP endpoint running a real sidecar against mock paid APIs; mock money, no login
- [Client SDK on GitHub](https://github.com/miraistudio/spendctl-sdk): Apache-2.0 thin clients
- [Japanese documentation](https://spendctl.dev/ja/docs/)