Add a spend firewall to your AI agent framework
TL;DR: Add a spend firewall to LangChain, CrewAI, OpenAI Agents SDK, or Vercel AI SDK. One call approves, blocks, or flags every agent purchase.
sipi.bot drops into every major agent framework with one call before a spend: approve, block, or flag against your rules. Pick your stack.
- LangChain agents
Add a spend firewall to your LangChain agent in 5 lines. Approve, block, or flag every tool-driven purchase be…
- CrewAI
Give your CrewAI crew a spend firewall as a BaseTool. Every agent purchase is approved, blocked, or flagged ag…
- the OpenAI Agents SDK
Wrap your OpenAI Agents SDK function tools with a spend firewall. Approve, block, or flag every purchase befor…
- the Vercel AI SDK
Add a spend firewall to your Vercel AI SDK tools in TypeScript. Approve, block, or flag every agent purchase b…
Every integration uses the same core call — evaluate(amount, merchant, category) returning APPROVED, BLOCKED, or FLAGGED — over a zero-dependency client. Open-source core (MIT), free to self-host, or $99/mo hosted with a live dashboard and tamper-evident audit log.
Frequently asked
Which agent frameworks does sipi.bot support?
sipi.bot works with LangChain, LangGraph, CrewAI, the OpenAI Agents SDK, and the Vercel AI SDK today, plus any runtime that can make an HTTP call or use an MCP tool.
Is there one client for all frameworks?
Yes. Python frameworks use the zero-dependency sipi_guard.py; TypeScript frameworks use sipiGuard.ts. Both wrap the same /v1/transactions/evaluate endpoint.