Stripe AI Developer Toolkit
Connect Stripe billing, token metering, MCP, and agent workflows to AI products.
What it does & when to use it
Stripe AI is a one-stop repository for building AI-powered products and businesses on Stripe. It includes @stripe/ai-sdk for connecting Stripe billing infrastructure to Vercel’s ai and @ai-sdk libraries, plus @stripe/token-meter for native OpenAI, Anthropic, and Google Gemini SDK integrations. Stripe also hosts a remote MCP server with OAuth-based client access and documents autonomous-agent use cases. The repository additionally provides Stripe agent skills and installation paths for several agent harnesses.
Provides @stripe/ai-sdk for integrating Stripe billing infrastructure with Vercel’s ai and @ai-sdk libraries; provides @stripe/token-meter for integrating Stripe billing infrastructure with native OpenAI, Anthropic, and Google Gemini SDKs; exposes the remote MCP server at https://mcp.stripe.com with OAuth-based client access; and documents installation commands for Claude Code, Codex, Cursor, and manual skills installation through npx skills add https://docs.stripe.com. Because SKILL.md is not cached, the source does not establish the concrete read, execution, or output behavior of an individual skill.
- A developer building a Stripe-connected application with Vercel’s AI libraries needs to add billing infrastructure.
- A team using the native OpenAI, Anthropic, or Google Gemini SDKs needs to meter token usage.
- A developer needs an MCP client to access Stripe capabilities through OAuth.
- A developer wants official Stripe agent tools inside Claude Code, Codex, or Cursor.
- A project team wants to install Stripe agent skills manually and manage their updates itself.
Pros & cons
- Covers both the Vercel AI ecosystem and multiple model vendors’ native SDKs.
- Includes components focused on Stripe billing infrastructure and token usage metering.
- Provides a remote MCP server with OAuth access.
- Offers installation paths for Claude Code, Codex, and Cursor, and uses the MIT license.
- The uncached SKILL.md prevents verification of the skill’s exact instructions, inputs, outputs, and runtime boundaries.
- The README provides no complete API examples, test results, or detailed platform validation.
- Manual installations require users to run the update command themselves.
- The source does not specify pricing, quotas, error handling, or the exact scope of MCP permissions.
How to install
From the project directory, use the command for the chosen agent harness: Claude Code: claude plugin install stripe@claude-plugins-official; Codex: codex plugin add stripe@openai-curated; Cursor: /add-plugin stripe. For manual installation, run npx skills add https://docs.stripe.com. The README does not provide a standalone SKILL.md path. Manually installed skills do not auto-update; run npx skills update -y to update them.
How to use
Choose @stripe/ai-sdk for Vercel’s ai and @ai-sdk libraries, or @stripe/token-meter for native OpenAI, Anthropic, and Google Gemini SDKs. For MCP access, connect a client to https://mcp.stripe.com and complete OAuth as described in Stripe’s documentation. The README does not provide a concrete API example, trigger prompt, or individual skill usage procedure.
Compared to similar skills
For Vercel’s ai and @ai-sdk libraries, the repository points to @stripe/ai-sdk; for native OpenAI, Anthropic, or Google Gemini SDK integrations, it points to @stripe/token-meter. The README presents these as separate integration paths but provides no performance or feature benchmark.