Install the package: `pip install paypal-agent-toolkit` (requires Python ≥ 3.11).
Set your PayPal client ID and secret as environment variables (`PAYPAL_CLIENT_ID`, `PAYPAL_CLIENT_SECRET`) and choose `sandbox` or `production` mode.
For LangChain: import `PayPalToolkit` from the package, instantiate it with your credentials, and call `.get_tools()` to retrieve a list of LangChain-compatible tools to pass to your agent executor.
For CrewAI: instantiate the toolkit and pass the tools list to your `Agent` definition via the `tools` parameter.
Use the built-in tools to create orders, capture payments, send invoices, and retrieve order details — each tool maps to a PayPal REST API operation.
Known gotchas
The package `paypal-agent-toolkit` is live on PyPI with adapters for LangChain, CrewAI, OpenAI Agents SDK, Amazon Bedrock, and Vercel AI SDK.
Sandbox and production credentials are separate; always test end-to-end in the PayPal sandbox before switching to production credentials.
Python ≥ 3.11 is required; the package will not install on older Python versions.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp