API & Integration Building

API Integration Architect Skill

Design secure, reliable API integrations — auth flows, endpoint mapping, request/response structures, error handling, and testing — without inventing endpoints or auth methods.

FreeChatGPTClaudeGeminiCopilotNotion AI
TL;DR

API Integration Architect Skill is a free AI skill for api & integration building. Design secure, reliable API integrations — auth flows, endpoint mapping, request/response structures, error handling, and testing — without inventing endpoints or auth methods. It works with ChatGPT, Claude, Gemini and is ready to use out of the box.

Best for
  • Planning a Stripe or payment provider integration
  • Wiring Slack, Discord, or Teams webhook notifications
  • Syncing customer data between a CRM and an app

About this skill

Most AI integration advice invents endpoints, guesses at auth methods, and skips the boring-but-critical parts like retries, rate limits, and webhook verification. This skill acts as an experienced integration architect: it clarifies the business goal, maps only confirmed endpoints, plans authentication against documented methods, and produces a full integration plan covering request/response shapes, error handling, testing, deployment, and maintenance. Every assumption is labeled, and unknowns become open questions instead of fabricated details.

What it does

Produces an API Integration Plan, Authentication Plan, Endpoint Mapping, Request/Response Structures, Error Handling Plan, Testing Checklist, and Deployment Checklist — grounded in confirmed documentation, with assumptions flagged explicitly.

Use cases

  • Planning a Stripe or payment provider integration
  • Wiring Slack, Discord, or Teams webhook notifications
  • Syncing customer data between a CRM and an app
  • Designing OAuth flows for third-party platform access
  • Adding webhook receivers with signature verification
  • Debugging an existing integration's failure modes
  • Scoping an integration before writing any code

What is included

  • SKILL.md
  • workflow.md
  • api-integration-framework.md
  • integration-checklist.md
  • output-templates.md
  • examples.md

How to use it

1. Download the ZIP using the Download button.
2. Unzip the api-integration-architect-skill folder.
3. Load SKILL.md into your AI assistant as a skill / system prompt.
4. Describe the systems you want to connect and the integration goal, and share any known API docs.

Examples

Example input
We need to integrate Stripe into our app to accept one-time payments.
Example output
Goal: Accept one-time card payments in our web app
Systems Involved: Web app ↔ Backend server ↔ Stripe API
Trigger: User submits checkout form
Data Flow Summary: Client requests a payment intent from backend; backend creates it via Stripe API; client confirms payment with returned client_secret using Stripe.js.

Authentication: API key (secret key server-side only, publishable key client-side). Confirmed by Stripe docs.
Credential Storage: Environment variables / secrets manager. Never in source.

Key Endpoints (verify current field names against Stripe docs at implementation time):
- POST /v1/payment_intents — create intent (server-side)
- GET  /v1/payment_intents/{id} — retrieve status
- POST {your webhook URL} — receive Stripe events (verify Stripe-Signature header)

Error Handling:
- 401 → re-check secret key, alert
- 400 → surface validation error to user
- 429 → exponential backoff + retry
- 5xx → retry with backoff, then alert
- Timeout → single retry, then fail-safe

Open Questions / Assumptions:
- Assumed web checkout, not mobile SDK — confirm platform.
- Assumed one-time payments only, no subscriptions — confirm.

Quality and safety notes

Quality checks
  • No invented endpoints, parameters, or response fields
  • Authentication method is confirmed against docs or flagged as an assumption
  • Every assumption is explicitly labeled
  • Least-privilege scopes recommended for all credentials
  • Retries, rate limits, and timeouts are addressed, not skipped
Safety notes
  • Never suggests hardcoding credentials — always environment variables or a secrets manager.
  • Recommends webhook signature verification for any incoming webhook.
  • Flags PII and payment data and recommends encryption in transit and at rest.
  • Recommends logging that excludes tokens, secrets, and sensitive payment details.

FAQ

Will it give me the exact Stripe / Slack / HubSpot endpoints?
Only when they are confirmed by your provided docs or well-known documented conventions. Otherwise it flags them as assumptions and asks you to confirm against current docs — never invents endpoints.
Does it write the integration code for me?
No. It produces the integration plan, auth plan, endpoint map, request/response shapes, and checklists. A developer implements from that plan.
Does it cover webhooks?
Yes — including signature verification, retry semantics, and idempotency for safe re-delivery.
Does it handle OAuth flows?
Yes. It plans OAuth 2.0 flows, scopes, token storage, and refresh handling when the provider supports OAuth.
Works with ChatGPT, Claude, Gemini, and Copilot?
Yes — it's model-agnostic and works with any general-purpose LLM assistant.