AgentPays
Wire governed payments into agent workflows so autonomous tools can spend within rules instead of sharing your main card.
Overview
AgentPays is a MCP server for the Build phase that gives AI agents governed spending via rules, approvals, and single-use virtual cards.
What is this MCP server?
- Spending rules and approval flows so agents cannot exceed budgets you define
- Single-use virtual cards for isolated checkout sessions
- Remote MCP at agentpays.dev with Bearer API key or OAuth on Claude Desktop, Claude.ai, and ChatGPT
- Streamable HTTP transport (MCP schema 2025-12-11), version 1.0.0
- Purpose-built payment infrastructure for AI agents, not generic card widgets
- MCP server version 1.0.0
- One streamable-http remote: https://agentpays.dev/api/mcp
- Schema: MCP server.schema.json 2025-12-11
What problem does it solve?
Letting coding agents pay for things without guardrails means surprise charges, shared cards, and no audit trail.
Who is it for?
Solo builders prototyping or shipping agents that need to autonomously pay for APIs, tools, or one-off purchases with caps and approvals.
Skip if: Teams that only need human checkout on a marketing site or a full merchant-of-record stack without any agent spend surface.
What do I get? / Deliverables
After you add the remote MCP, your agent can initiate purchases inside spending rules and approval flows instead of bypassing your wallet.
- Remote MCP connection to governed agent payment APIs
- Ability to issue single-use virtual cards under configured rules
- Approval-gated spend paths the agent can invoke from the IDE
Recommended MCP Servers
Journey fit
Payment MCPs land in Build because you integrate them while wiring agent tooling, backends, and third-party APIs before production hardening. Integrations is the right shelf for streamable HTTP remotes that connect Claude, Cursor, or Codex to external financial APIs during implementation.
How it compares
Agent-governed spend MCP, not a storefront payment skill or generic Stripe checkout tutorial.
Common Questions / FAQ
Who is dev.agentpays/agentpays for?
Indie and solo builders who run Claude Code, Cursor, or similar agents and need safe, policy-bound purchasing during product development.
When should I use dev.agentpays/agentpays?
Use it while building agent integrations whenever the agent must pay vendors under limits, not when you are only validating ideas without real spend.
How do I add dev.agentpays/agentpays to my agent?
Register the streamable HTTP remote https://agentpays.dev/api/mcp with Authorization Bearer your agent API key, or use OAuth where the host supports it.