# Regent Protocol

> Regent decides, before money moves, whether an AI agent may do what it is about to do. A person signs a mandate (limits and scope), every payment the agent attempts is decided against it and the organization's policy, what exceeds the limits is escalated to a person, and every decision leaves a signed receipt anyone can verify without trusting Regent. Works on any payment rail: Stripe, the owner's own card through Agentcard, co-signed custody, stablecoins. Live today; Solana devnet is one of the audit witnesses.

## If you are an agent

Read https://docs.regentprotocol.org/agent.md first: the whole path in order (pick your door, get your credential, ask `my_budget` what comes next and say it to the owner, buy, show the receipt). The MCP server is `https://gw.regentprotocol.org/gw/mcp` (Streamable HTTP, OAuth 2.1).

Three doors, one set of rules (you act only as yourself, you never hold an organization key, a person grants every mandate):

- You run inside an app with custom MCP connectors (Claude, Claude Code, ChatGPT, Grok Bot): the person adds `https://gw.regentprotocol.org/gw/mcp` as a connector and signs in. https://docs.regentprotocol.org/gateway/connect-from-your-app.md
- You run on a machine with a terminal: `pip install regent-control` and `regent-control enroll <one-time code from the console>`. https://docs.regentprotocol.org/control/enrol-agent.md
- You can only send HTTP with a stored secret: the person gives you a credential bound to you from the consent screen. https://docs.regentprotocol.org/gateway/tools-mcp.md

Start with the agent skill: https://docs.regentprotocol.org/skill.md

## Pages

- [Home](https://regentprotocol.org/): what Regent does — the financial action flow from AgentID through mandate check, Guardian decision, and audit proof
- [Technology](https://regentprotocol.org/technology): architecture — Solana settlement programs (devnet), Merkle-anchored audit evidence, Guardian risk engine, Python SDK
- [Pricing](https://regentprotocol.org/pricing): usage-based launch pricing — Developer $0, Growth $299/mo, Scale $999/mo, Enterprise from $2,500/mo; the unit is a verified action, never seats
- [Pilot Program](https://regentprotocol.org/pilot): structured 14-day engagement; application form online
- [About](https://regentprotocol.org/about): team and thesis
- [Terms of Use](https://regentprotocol.org/terms) · [Privacy Policy](https://regentprotocol.org/privacy)

## Developer resources

- [Documentation](https://docs.regentprotocol.org): connect from your agent app, enrol an agent, mandates and approvals, REST API, Python SDK (`pip install regent-control`), Solana programs — machine-readable index at [docs llms.txt](https://docs.regentprotocol.org/llms.txt), skill at [skill.md](https://docs.regentprotocol.org/skill.md)
- [Verify on-chain](https://docs.regentprotocol.org/solana-programs/verification): independently check agents, mandates, and audit events against Solana

## Contact

info@regentprotocol.org · [Pilot application](https://regentprotocol.org/pilot)