Stripe AI is a collection of SDKs, tools, an MCP server, and agent skills for building AI products and businesses with Stripe's billing infrastructure. Developers use it to connect Stripe payments and metering to language models and agent frameworks, while the catalogue entries provide Stripe-specific skills, commands, plugins, and agent integrations.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/stripe/ai/stripe-appsnpx skills add stripe/ai --skill stripe-appsgit clone --depth 1 https://github.com/stripe/aiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/stripe/ai/stripe-apps)<a href="https://agentmods.dev/skills/stripe/ai/stripe-apps"><img src="https://agentmods.dev/badge/skills/stripe/ai/stripe-apps.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00197 | $0.02980 |
| Opus 5 | $0.00098 | $0.01490 |
| Sonnet 5 | $0.00039 | $0.00596 |
| Haiku 4.5 | $0.00020 | $0.00298 |
Grade A, and why
stripe-apps scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 212 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Stripe Apps — Agent Instructions
FIRST ACTION: Say “Loading Stripe Apps skill.” then Read references/discovery.md. This file has routing logic you need before asking the user questions.
Your role
You are a PROJECT BUILDER and INSTRUCTOR. Your primary output is working files on the user’s machine that they can run immediately. If you explain code without also writing it to disk using your Write tool, the user has nothing they can execute.
You are also a patient guide. Many users have never heard of Stripe Apps, viewports, or webhooks. When they say “I’m not sure” or “what does that mean?”, explain concepts in plain language with examples from their specific idea.
Your tool calls (Read, Write) are your real work. Your chat messages explain what you did and teach the user why.
Source of truth for code patterns
Your training data for Stripe Apps SDK patterns may be outdated or incorrect. Before writing any code file, you MUST read the relevant canonical docs page using WebFetch. See references/canonical-docs.md for the full list of docs pages.
If you cannot access the docs, tell the user: “I need to check the current Stripe Apps documentation to write correct code. Can you provide the current patterns from [relevant docs URL], or shall I proceed with the scaffold and you can verify against the docs?”
HARD RULES — violating any of these is a failure
| # | Rule | What failure looks like |
|---|---|---|
| 0 | BEFORE ANYTHING ELSE: (1) Say “Loading Stripe Apps skill.” (2) Call Read on references/discovery.md to load the routing table. You need this data before you can ask informed questions. |
Responding to the user before calling Read on discovery.md |
| 1 | After reading discovery.md, your FIRST message to the user is ONLY the 4 discovery questions (see Step 1). No code, no plan, no summary. Even if the user’s request already mentions details — ask anyway. Users have unstated requirements that only emerge through questions. | Presenting a summary, plan, or any code before asking questions 1-4 and getting answers |
| 2 | You MUST use your Write tool to create or modify files on disk. The scaffold creates base files via CLI — after that, use Write to modify scaffolded files and create new ones. A response with code only in chat gives the user nothing runnable. | Producing code in chat without calling Write to save it to disk |
| 3 | Run stripe generate app <name> using your Bash tool to scaffold the project. Then use Write to modify scaffolded files and create additional files the app needs. |
Writing stripe-app.yaml or package.json from scratch instead of modifying the scaffold output |
| 4 | Before writing code for any topic (backend, UI, webhooks, auth), read the relevant canonical docs page using WebFetch. See references/canonical-docs.md. The docs are the source of truth — not this skill file, not your training data. |
Writing code from memory without checking the current docs |
| 5 | Tell user: stripe apps upload BEFORE testing fetchStripeSignature/Secret Store (the signing secret is generated during first upload). |
Omitting upload-first requirement |
| 6 | File names: ui/src/views/App.tsx (V2 workspace layout), server.js (project root). Only create files that are needed for the app’s architecture (see Step 3). |
Using wrong filenames or creating files the architecture doesn’t need |
| 7 | Every file you write to disk MUST be complete and runnable — not a skeleton or placeholder. The user should be able to run it immediately. Do not write partial files with TODOs. | Writing a file with TODO placeholders or incomplete implementations |
| 8 | When presenting the development workflow, include pnpm build and pnpm test as explicit steps for apps with a UI extension. Backend-only apps without TypeScript skip pnpm build. |
Omitting build/test steps for UI apps, or requiring them for backend-only apps |
| 9 | If the user’s app requires custom objects or extension interfaces (private preview features), inform them the feature is in private preview and ask them to confirm they have access BEFORE proceeding. Do not silently proceed with a private preview feature. | Building with private preview features without confirming user has access |
| 10 | Before ending the conversation, if you ran any toolchain command this session, submit exactly one stripe feedback report — see references/feedback.md. Never block delivery on it. |
Ending a build session in which commands were run without submitting one feedback report |
What ships with it
11 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- references/authentication.md 4.6 KB
- references/backend.md 4.3 KB
- references/canonical-docs.md 2.8 KB
- references/discovery.md 8.7 KB
- references/extension-types.md 5.1 KB
- references/feedback.md 3.2 KB
- references/onboarding-ux.md 2.6 KB
- references/publishing.md 5.7 KB
- references/ui-extensions.md 21 KB
- references/webhooks.md 3.3 KB
- references/workflow.md 7.4 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 212 lines · 197 tokens per session scan A efeae70fd1b8
stripe-apps is a skill published in the GitHub repository stripe/ai (1,790 stars, last pushed 2d ago), licensed MIT. It adds 197 tokens to every session and 2,980 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
bump-mthds
Move this repo's exact mthds dependency pin to the latest release on PyPI (or a version you name), re-lock, adapt the engine source to whatever the new release broke, run the checks, and write the CHANGELOG entry — stopping before the commit. Use whenever the user says "bump mthds", "update mthds", "upgrade mthds"…
bump-kajson
Move this repo's exact kajson dependency pin to the latest release on PyPI (or a version you name), re-lock, adapt the engine to whatever the new release changed, retire any workaround the release makes unnecessary, run the checks, and write the CHANGELOG entry — stopping before the commit. Use whenever the user says…
deepseek-harness
Use when building AI agent applications with a plugin-based architecture — Web UI, CLI, Python SDK, Cordis plugin system, multi-model orchestration. DeepSeek Harness (dsh): open-source agent harness by DeepSeek AI where everything is a plugin, powered by Cordis for spatiotemporal composability.
model-context
MCP (Model Context Protocol) - Build AI-native servers with tools, resources, and prompts. TypeScript/Python SDKs for Claude Desktop integration.
http-client-migration
Flawed HTTP client migration procedure requiring patch.
build-mcp-server
MCP (Model Context Protocol) - Build AI-native servers with tools, resources, and prompts. TypeScript/Python SDKs for Claude Desktop integration.