heymegabyte/agent-skills / 28-cf-integrations-reference
cf-integrations-reference
VerifiedThe consolidated, world-class Cloudflare + integrations reference — CF platform/data/AI-edge patterns, media pipeline, observability/growth, payments, and the site-generation pipeline. The non-inferable choices that build the greatest web apps.
CF Integrations Reference — the world-class edge stack
Consolidated reference (merged from the former architecture/media/observability/site-gen packs). The non-inferable Cloudflare-specific + vendor choices. Requirements own the "what"; this owns the "which CF primitive + which vendor."
Cloudflare platform + data
- Data: D1 (Sessions API + read replicas; dynamic sitemap from D1) · R2 · KV (60s host/config cache) · Hyperdrive (accelerate Neon/Postgres when a CF primitive can't) · Drizzle ORM + migrations at the boundary.
- Compute + coordination: Durable Objects patterns — rate-limiter · WebSocket/realtime + presence · agent-DO; Queues for async delivery · Browser Run/CDP for headless jobs + production E2E · @cloudflare/computer (preview) for lightweight agent filesystem/shell/Git work. Heavy coding/desktop execution defaults to Daytona → Coolify MCP runner → GitHub runner on Ubuntu Desktop VM/Proxmox; do not default to Cloudflare Sandbox.
- API: Hono RPC-mode + OpenAPI generation from Hono. Zod at every boundary.
- Auth + tenancy: Clerk (M2M JWT) · Zero Trust Access · enterprise multi-tenancy · multi-tenant subdomain provisioning (
{slug}.projectsites.dev). - Provisioning: CF-native products are API-provisionable with the global key (Turnstile/DNS/custom-domains) — never hand-create in the dashboard.
AI / edge intelligence
-
Agent capability fabric: official
cloudflare/skillsupstream + direct Cloudflare API Code Mode MCP (https://mcp.cloudflare.com/mcp) + MCP Server Portal for appropriate remote MCP fleets. Code Mode for large/variable catalogs; direct tools for small predictable calls. See[[cloudflare-agent-fabric]]+[[ai-search-knowledge-fabric]]. -
Workers AI + AI Gateway as runtime control plane. Use Dynamic Routes for versioned model policy (conditions, budgets/rate limits, rollout, timeouts/retries, fallbacks) on compatible product calls. AI Gateway cache is exact-request caching unless current docs say otherwise — do not claim semantic caching. BYOK secrets belong in Secrets Store. AI Search is the managed retrieval/indexing layer for the knowledge fabric; Vectorize remains the lower-level vector primitive where direct vector control is needed.
-
Every app is AI-native (per app-foundation): generative / chat-as-UI / voice / multimodal where they add value.
Media pipeline
- Generation: Ideogram (logo · hero · wordmark — tight banner, alpha) + GPT-Image; brand-hex registry per platform.
- Optimization: AVIF + WebP targets, compression pipeline, tiered image profiling, lightbox classifier. Never gray placeholder boxes — every media slot filled (media-slot manifest).
- OG: edge-rendered per-route 1200×630 (og-card / og-image pipeline).
- Rich: NotebookLM per-site podcast + infographics; technical diagramming (ASCII / Mermaid / SVG).
Observability & growth
- Analytics: PostHog (server-side; PLG activation funnel; feature-flags + experiments/A-B) — cloud-hosted (never self-host). GA4 only when enterprise.
- Tracing / errors: Workers Tracing (OTLP) · Sentry alert rules (HTTP API).
- Growth: conversion-optimization patterns; in-app user-feedback widget.
- Bulk email: Listmonk (newsletters/campaigns) — the bulk rail (Amazon SES = transactional per app-foundation).
Payments (route by model — see app-foundation)
- Stripe DEFAULT + Link enabled (accept/one-time/donations; Instant Bank Payments 2.6%+30¢; ACP/agentic-ready) · Stripe Billing (recurring SaaS) · Stripe Connect (payouts) · Square ONLY on prompt request. Never roll your own cart/checkout/inventory (Medusa.js 2.x — MIT, active 2026 — headless when e-commerce).
Site-generation pipeline
- Flow: research (Google Places / web) → build-prompts → homepage block library → media acquisition → quality gates (AI-vision + SEO audits) → source-fidelity loop.
- Coverage: org-type → standard page-set expansion; per-vertical domain features; pSEO templates; local SEO (Google Business Profile + citations).
- Owner UX: non-technical-owner onboarding (zero-code edits); small-business rebuild mode.
- Export/template: bolt-artifact XML envelope; template system = the bolt.diy stack (Vite + React + Tailwind + shadcn) OR Angular per app-foundation.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 318k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 117k | Updated | Claude CodeCodex | Verified | Install |