Cloudflare Wrangler Deployment Guide
Helps agents reliably build, develop, deploy, and manage Cloudflare Workers with Wrangler.
The skill prefers official documentation and the configuration schema, and includes partial safeguards for secrets, remote bindings, logs, and deletion. However, deployment, deletion, remote database, and billable AI operations lack explicit per-action user confirmation, least-privilege guidance, data-flow disclosure, and preflight recovery checks, so points are deducted.
The material is broad, internally organized, and maps commands to troubleshooting guidance, including local versus remote modes. However, no key-path test suite or reproducible static evidence is provided; version drift is acknowledged but not controlled, and failure diagnostics remain thin, so the score is conservatively limited below the static ceiling.
The audience and major scenarios are reasonably clear, covering development, configuration, deployment, and resource management, with retrieval-first guidance. Boundaries, non-fit cases, input/output contracts, and Chinese-language support are not specified; core retrieval points to Cloudflare-hosted documentation whose mainland-China reachability is unverified, so environment fit is only partial.
The document has useful progressive sections, quick-start material, command tables, configuration examples, testing, and troubleshooting. It lacks skill-specific versioning, changelog, named maintenance responsibility, a clear update path, dependency version policy, and fuller limitations or FAQs. Apache-2.0 licensing is present, but publisher identity is not verified.
The skill supplies directly reusable commands and configuration snippets for common Wrangler workflows and covers the core task categories. Static review provides no execution evidence for command correctness; several examples depend on placeholders, account state, and external resources, and success criteria and output contracts are limited, so the score remains below the static maximum.
Official documentation, schema, and Cloudflare documentation are named as traceability sources, and the repository includes a Semgrep CI workflow. There is no committed test coverage for Wrangler key paths, execution evidence, or meaningful cross-source corroboration; Semgrep does not establish command executability, so only limited credit is warranted.
- Do not run deploy, delete, rollback, remote database, or remote AI operations before confirming the target account, environment, resources, and billing impact.
- Remote bindings and Workers AI may access real data and incur charges; verify configuration, permissions, account scope, and data scope first.
- Commands, configuration fields, and the 2026-01-01 compatibility-date example may age; validate against current official documentation and the local schema before use.
- No key-command tests or execution records are provided, so examples should not be treated as verified procedures.
- Mainland-China reachability of Cloudflare documentation and services is not established by the supplied material.
What does this skill do, and when should you use it?
This Agent Skill provides operational guidance for the Wrangler CLI in Cloudflare Workers projects. It tells agents to retrieve current Cloudflare documentation and configuration schemas instead of relying on potentially stale knowledge. The scope includes Worker initialization, local development, deployment, configuration, type generation, logs, testing, and management of KV, R2, D1, Vectorize, Hyperdrive, Workers AI, Queues, Containers, Workflows, Pipelines, and Secrets Store. It fits developers who manage Cloudflare projects from the command line, while execution, authentication, and resource permissions still depend on the local environment.
It instructs the agent to check wrangler --version and install wrangler@latest when needed; configure Workers, environments, compatibility dates, and resource bindings in wrangler.jsonc; run commands such as wrangler dev, wrangler deploy, wrangler types, wrangler tail, and wrangler check startup; manage KV, R2, D1, Vectorize, Hyperdrive, Workers AI, Queues, Containers, Workflows, Pipelines, Secrets Store, and Pages; and handle local secrets, remote bindings, deployment rollback, Vitest testing, and common troubleshooting scenarios.
- A Cloudflare Workers developer needs to initialize a Worker and create a `wrangler.jsonc` configuration.
- A team needs to test a Worker locally with simulated bindings or selected remote resources.
- An engineer needs to deploy a Worker, generate TypeScript binding types, inspect live logs, or profile startup time.
- A developer needs Wrangler commands for creating and managing KV, R2, D1, Vectorize, Queues, or other listed Cloudflare resources.
- A team needs separate staging and production environments and safer handling of deployment secrets.
What are this skill's strengths and limitations?
- Covers Wrangler workflows for development, deployment, configuration, debugging, and resource management.
- Explicitly prioritizes current documentation, reducing the risk of stale CLI flags or configuration guidance.
- Includes practical workflows for bindings, environments, type generation, secrets, and rollback.
- Provides concrete security guidance for handling secret values.
- Its guidance depends on Cloudflare documentation, the configuration schema, and the installed Wrangler version, so examples may require updates.
- It requires shell, filesystem, and network access plus Node.js and Wrangler.
- The source provides no independent test suite or platform compatibility results.
- Workers AI runs remotely and incurs usage charges even during local development.
How do you install this skill?
The repository is a collection of multiple skills. Install the collection with npx skills add https://github.com/cloudflare/skills; to use only this skill, copy the repository's skills/wrangler folder into the skill directory for your agent. The README lists ~/.claude/skills/, ~/.cursor/skills/, ~/.config/opencode/skills/, ~/.codex/skills/, and ~/.pi/agent/skills/ for Claude Code, Cursor, OpenCode, OpenAI Codex, and Pi respectively. The source does not document a dedicated single-skill installation command.
How do you use this skill?
Invoke it when a task involves Wrangler commands or Cloudflare Workers configuration, for example: “Use Wrangler to initialize a Worker, configure D1 and R2, and provide local development and deployment steps.” The agent should first retrieve current information from https://developers.cloudflare.com/workers/wrangler/ and related sources, then check whether Wrangler is installed. If it is missing, run npm install -D wrangler@latest, followed by commands such as wrangler dev, wrangler deploy, or wrangler types as appropriate.