Generated banner for Cloudflare Deploy
Generated logo for Cloudflare Deploy

Cloudflare Deploy

OpenAIOfficialPublished by OpenAI

  • Skill
  • Tool integration
  • Cloud and DevOps
  • Apache-2.0
  • Medium risk

From the author

Deploy applications and infrastructure to Cloudflare using Workers, Pages, and related platform services. Use when the user asks to deploy, host, publish, or set up a project on Cloudflare.
— the description by OpenAI

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

Cloudflare Deploy provides a centralized framework for developers to build, host, and manage projects on Cloudflare. It covers the entire ecosystem, including serverless compute (Workers), full-stack web hosting (Pages), persistent storage (D1, R2, KV), and AI-powered agent development. The skill includes decision trees to help select the right product for specific needs, authentication guidance, and troubleshooting steps for deployment issues.

What it does

  • Provides decision trees to select the appropriate Cloudflare product for compute, storage, AI, or networking.
  • Offers guidance on authenticating with the Cloudflare platform via CLI or CI/CD.
  • Includes a comprehensive product index for quick reference to documentation.
  • Provides specific patterns and API references for building stateful AI agents using the Agents SDK.
  • Offers troubleshooting steps for network-related deployment failures.
  • Supports infrastructure-as-code integration with tools like Pulumi and Terraform.

When to use it

  • When you need to deploy, host, or publish a project to Cloudflare.
  • When you are unsure which Cloudflare service (e.g., Workers vs. Pages) best fits your application requirements.
  • When you need to build stateful, globally distributed AI agents.
  • When you encounter network errors or timeouts during the deployment process.
  • When setting up persistent storage or database solutions within the Cloudflare ecosystem.

Advantages

  • Consolidates fragmented documentation into a single, navigable resource.
  • Simplifies product selection through clear, use-case-driven decision trees.
  • Provides actionable code patterns for complex tasks like AI agent development.
  • Addresses common deployment friction points like authentication and network sandboxing.

Usage examples

“I need to deploy a full-stack web application to Cloudflare. Which service should I use?”
The agent will guide you to use Cloudflare Pages for full-stack web apps with Git-based deployments.
“How do I build an AI chat agent on Cloudflare?”
The agent will provide the Agents SDK reference, including the AIChatAgent class and examples for streaming responses and managing message history.
“My deployment is failing with a network timeout error. What should I do?”
The agent will advise you to rerun the deployment command with the sandbox_permissions=require_escalated flag to bypass network restrictions.
“How do I store relational data for my worker?”
The agent will recommend using D1 for SQLite-based relational storage or Hyperdrive for connecting to existing Postgres/MySQL databases.

Install

  1. Download the skill as a ZIP:

    https://autogineer.com/agent-skills/openai/cloudflare-deploy/download
  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip cloudflare-deploy.zip -d .agents/skills

Documentation by OpenAI

Cloudflare Deploy

Consolidated skill for building on the Cloudflare platform. Use decision trees below to find the right product, then load detailed references.

Prerequisites

  • When sandboxing blocks the deployment network calls, rerun with sandbox_permissions=require_escalated.

  • The deployment might take a few minutes. Use appropriate timeout values.

Authentication (Required Before Deploy)

Verify auth before wrangler deploy, wrangler pages deploy, or npm run deploy:

npx wrangler whoami    # Shows account if authenticated

Not authenticated? → references/wrangler/auth.md

  • Interactive/local: wrangler login (one-time OAuth)

  • CI/CD: Set CLOUDFLARE_API_TOKEN env var

Quick Decision Trees

"I need to run code"
Need to run code?
├─ Serverless functions at the edge → workers/
├─ Full-stack web app with Git deploys → pages/
├─ Stateful coordination/real-time → durable-objects/
├─ Long-running multi-step jobs → workflows/
├─ Run containers → containers/
├─ Multi-tenant (customers deploy code) → workers-for-platforms/
├─ Scheduled tasks (cron) → cron-triggers/
├─ Lightweight edge logic (modify HTTP) → snippets/
├─ Process Worker execution events (logs/observability) → tail-workers/
└─ Optimize latency to backend infrastructure → smart-placement/
"I need to store data"
Need storage?
├─ Key-value (config, sessions, cache) → kv/
├─ Relational SQL → d1/ (SQLite) or hyperdrive/ (existing Postgres/MySQL)
├─ Object/file storage (S3-compatible) → r2/
├─ Message queue (async processing) → queues/
├─ Vector embeddings (AI/semantic search) → vectorize/
├─ Strongly-consistent per-entity state → durable-objects/ (DO storage)
├─ Secrets management → secrets-store/
├─ Streaming ETL to R2 → pipelines/
└─ Persistent cache (long-term retention) → cache-reserve/
"I need AI/ML"
Need AI?
├─ Run inference (LLMs, embeddings, images) → workers-ai/
├─ Vector database for RAG/search → vectorize/
├─ Build stateful AI agents → agents-sdk/
├─ Gateway for any AI provider (caching, routing) → ai-gateway/
└─ AI-powered search widget → ai-search/
"I need networking/connectivity"
Need networking?
├─ Expose local service to internet → tunnel/
├─ TCP/UDP proxy (non-HTTP) → spectrum/
├─ WebRTC TURN server → turn/
├─ Private network connectivity → network-interconnect/
├─ Optimize routing → argo-smart-routing/
├─ Optimize latency to backend (not user) → smart-placement/
└─ Real-time video/audio → realtimekit/ or realtime-sfu/
"I need security"
Need security?
├─ Web Application Firewall → waf/
├─ DDoS protection → ddos/
├─ Bot detection/management → bot-management/
├─ API protection → api-shield/
├─ CAPTCHA alternative → turnstile/
└─ Credential leak detection → waf/ (managed ruleset)
"I need media/content"
Need media?
├─ Image optimization/transformation → images/
├─ Video streaming/encoding → stream/
├─ Browser automation/screenshots → browser-rendering/
└─ Third-party script management → zaraz/
"I need infrastructure-as-code"
Need IaC? → pulumi/ (Pulumi), terraform/ (Terraform), or api/ (REST API)

Product Index

Compute & Runtime
ProductReference
Workersreferences/workers/
Pagesreferences/pages/
Pages Functionsreferences/pages-functions/
Durable Objectsreferences/durable-objects/
Workflowsreferences/workflows/
Containersreferences/containers/
Workers for Platformsreferences/workers-for-platforms/
Cron Triggersreferences/cron-triggers/
Tail Workersreferences/tail-workers/
Snippetsreferences/snippets/
Smart Placementreferences/smart-placement/
Storage & Data
ProductReference
KVreferences/kv/
D1references/d1/
R2references/r2/
Queuesreferences/queues/
Hyperdrivereferences/hyperdrive/
DO Storagereferences/do-storage/
Secrets Storereferences/secrets-store/
Pipelinesreferences/pipelines/
R2 Data Catalogreferences/r2-data-catalog/
R2 SQLreferences/r2-sql/
AI & Machine Learning
ProductReference
Workers AIreferences/workers-ai/
Vectorizereferences/vectorize/
Agents SDKreferences/agents-sdk/
AI Gatewayreferences/ai-gateway/
AI Searchreferences/ai-search/
Networking & Connectivity
ProductReference
Tunnelreferences/tunnel/
Spectrumreferences/spectrum/
TURNreferences/turn/
Network Interconnectreferences/network-interconnect/
Argo Smart Routingreferences/argo-smart-routing/
Workers VPCreferences/workers-vpc/
Security
ProductReference
WAFreferences/waf/
DDoS Protectionreferences/ddos/
Bot Managementreferences/bot-management/
API Shieldreferences/api-shield/
Turnstilereferences/turnstile/
Media & Content
ProductReference
Imagesreferences/images/
Streamreferences/stream/
Browser Renderingreferences/browser-rendering/
Zarazreferences/zaraz/
Real-Time Communication
ProductReference
RealtimeKitreferences/realtimekit/
Realtime SFUreferences/realtime-sfu/
Developer Tools
ProductReference
Wranglerreferences/wrangler/
Miniflarereferences/miniflare/
C3references/c3/
Observabilityreferences/observability/
Analytics Enginereferences/analytics-engine/
Web Analyticsreferences/web-analytics/
Sandboxreferences/sandbox/
Workerdreferences/workerd/
Workers Playgroundreferences/workers-playground/
Infrastructure as Code
ProductReference
Pulumireferences/pulumi/
Terraformreferences/terraform/
APIreferences/api/
Other Services
ProductReference
Email Routingreferences/email-routing/
Email Workersreferences/email-workers/
Static Assetsreferences/static-assets/
Bindingsreferences/bindings/
Cache Reservereferences/cache-reserve/

Troubleshooting

Escalated Network Access

If deployment fails due to network issues (timeouts, DNS errors, connection resets), rerun the deploy with escalated permissions (use sandbox_permissions=require_escalated). The deploy requires escalated network access when sandbox networking blocks outbound requests.

Example guidance to the user:

The deploy needs escalated network access to deploy to Cloudflare. I can rerun the command with escalated permissions—want me to proceed?

Files

references/queues/gotchas.md

Queues Gotchas & Troubleshooting

CRITICAL: Top Production Mistakes

1. "Entire Batch Retried After Single Error"

Problem: Throwing uncaught error in queue handler retries the entire batch, not just the failed message
Cause: Uncaught exceptions propagate to the runtime, triggering batch-level retry
Solution: Always wrap individual message processing in try/catch and call msg.retry() explicitly

// ❌ BAD: Throws error, retries entire batch
async queue(batch: MessageBatch): Promise<void> {
  for (const msg of batch.messages) {
    await riskyOperation(msg.body); // If this throws, entire batch retries
    msg.ack();
  }
}

// ✅ GOOD: Catch per message, handle individually
async queue(batch: MessageBatch): Promise<void> {
  for (const msg of batch.messages) {
    try {
      await riskyOperation(msg.body);
      msg.ack();
    } catch (error) {
      msg.retry({ delaySeconds: 60 });
    }
  }
}
2. "Messages Retry Forever"

Problem: Messages not explicitly ack'd or retry'd will auto-retry indefinitely
Cause: Runtime default behavior retries unhandled messages until max_retries reached
Solution: Always call msg.ack() or msg.retry() for each message. Never leave messages unhandled.

// ❌ BAD: Skipped messages auto-retry forever
async queue(batch: MessageBatch): Promise<void> {
  for (const msg of batch.messages) {
    if (shouldProcess(msg.body)) {
      await process(msg.body);
      msg.ack();
    }
    // Missing: msg.ack() for skipped messages - they will retry!
  }
}

// ✅ GOOD: Explicitly handle all messages
async queue(batch: MessageBatch): Promise<void> {
  for (const msg of batch.messages) {
    if (shouldProcess(msg.body)) {
      await process(msg.body);
      msg.ack();
    } else {
      msg.ack(); // Explicitly ack even if not processing
    }
  }
}

Common Errors

"Duplicate Message Processing"

Problem: Same message processed multiple times
Cause: At-least-once delivery guarantee means duplicates are possible during retries
Solution: Design consumers to be idempotent by tracking processed message IDs in KV with expiration TTL

async queue(batch: MessageBatch, env: Env): Promise<void> {
  for (const msg of batch.messages) {
    const processed = await env.PROCESSED_KV.get(msg.id);
    if (processed) {
      msg.ack();
      continue;
    }
    
    await processMessage(msg.body);
    await env.PROCESSED_KV.put(msg.id, '1', { expirationTtl: 86400 });
    msg.ack();
  }
}
"Pull Consumer Can't Decode Messages"

Problem: Pull consumer or dashboard shows unreadable message bodies
Cause: Messages sent with v8 content type are only decodable by Workers push consumers
Solution: Use json content type for pull consumers or dashboard visibility

// Use json for pull consumers
await env.MY_QUEUE.send(data, { contentType: 'json' });

// Use v8 only for push consumers with complex JS types
await env.MY_QUEUE.send({ date: new Date(), tags: new Set() }, { contentType: 'v8' });
"Messages Not Being Delivered"

Problem: Messages sent but consumer not processing
Cause: Queue paused, consumer not configured, or consumer errors
Solution: Check queue status with wrangler queues list, verify consumer configured with wrangler queues consumer add, and check logs with wrangler tail

"High Dead Letter Queue Rate"

Problem: Many messages ending up in DLQ
Cause: Consumer repeatedly failing to process messages after max retries
Solution: Review consumer error logs, check external dependency availability, verify message format matches expectations, or increase retry delay

Error Classification Patterns

Classify errors to decide whether to retry or DLQ:

async queue(batch: MessageBatch, env: Env): Promise<void> {
  for (const msg of batch.messages) {
    try {
      await processMessage(msg.body);
      msg.ack();
    } catch (error) {
      // Transient errors: retry with backoff
      if (isRetryable(error)) {
        const delay = Math.min(30 * (2 ** msg.attempts), 43200);
        msg.retry({ delaySeconds: delay });
      } 
      // Permanent errors: ack to avoid infinite retries
      else {
        console.error('Permanent error, sending to DLQ:', error);
        await env.ERROR_LOG.put(msg.id, JSON.stringify({ msg: msg.body, error: String(error) }));
        msg.ack(); // Prevent further retries
      }
    }
  }
}

function isRetryable(error: unknown): boolean {
  if (error instanceof Response) {
    // Retry: rate limits, timeouts, server errors
    return error.status === 429 || error.status >= 500;
  }
  if (error instanceof Error) {
    // Don't retry: validation, auth, not found
    return !error.message.includes('validation') && 
           !error.message.includes('unauthorized') &&
           !error.message.includes('not found');
  }
  return false; // Unknown errors don't retry
}
"CPU Time Exceeded in Consumer"

Problem: Consumer fails with CPU time limit exceeded
Cause: Consumer processing exceeding 30s default CPU time limit
Solution: Increase CPU limit in wrangler.jsonc: { "limits": { "cpu_ms": 300000 } } (5 minutes max)

Content Type Decision Guide

When to use each content type:

Content TypeUse WhenReadable BySupports
json (default)Pull consumers, dashboard visibility, simple objectsAll (push/pull/dashboard)JSON-serializable types only
v8Push consumers only, complex JS objectsPush consumers onlyDate, Map, Set, BigInt, typed arrays
textString-only payloadsAllStrings only
bytesBinary data (images, files)AllArrayBuffer, Uint8Array

Decision tree:

  1. Need to view in dashboard or use pull consumer? → Use json

  2. Need Date, Map, Set, or other V8 types? → Use v8 (push consumers only)

  3. Just strings? → Use text

  4. Binary data? → Use bytes

// Dashboard/pull: use json
await env.QUEUE.send({ id: 123, name: 'test' }, { contentType: 'json' });

// Complex JS types (push only): use v8
await env.QUEUE.send({ 
  created: new Date(), 
  tags: new Set(['a', 'b']) 
}, { contentType: 'v8' });

Limits

LimitValueNotes
Max queues10,000Per account
Message size128 KBMaximum per message
Batch size (consumer)100 messagesMaximum messages per batch
Batch size (sendBatch)100 msgs or 256 KBWhichever limit reached first
Throughput5,000 msgs/secPer queue
Retention4-14 daysConfigurable retention period
Max backlog25 GBMaximum queue backlog size
Max delay12 hours (43,200s)Maximum message delay
Max retries100Maximum retry attempts
CPU time default30sPer consumer invocation
CPU time max300s (5 min)Configurable via limits.cpu_ms
Operations per message3 (write + read + delete)Base cost per message
Pricing$0.40 per 1M operationsAfter 1M free operations
Message chargingPer 64 KB chunkMessages charged in 64 KB increments

Requirements and compatibility

Written for: OpenAI Codex

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Needs Node.js

Security

What our scanner noticed· Medium risk

  • Possible data-exfiltration endpoints
  • Instructions that use the network
  • Downloads piped into a shell

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Published by
OpenAI
Publisher
OpenAI (Official)
Source
openai/skills · skills/.curated/cloudflare-deploy · commit 49f948f
Licence
Apache License 2.0 (redistribution allowed)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by OpenAI or OpenAI.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download cloudflare-deploy.zip