Cloudflare cf CLI

Finds and runs commands of Cloudflare's beta cf CLI through cf cli search, cf schema and dry runs, and picks cf or Wrangler per project. Use when managing DNS, zones, D1, KV or R2 from the terminal or starting a new Workers project.

DevOpsPack: Cloudflare skills94 linescf-cli
Author
Source
cloudflare/skills/skills/cf-cli @ 5108325
License
Apache-2.0

About this skill

What it does

The skill teaches the agent Cloudflare's cf CLI, which is in beta: more than 2,900 commands generated mostly from the public API, plus cf dev, cf build and cf deploy for Workers projects. Instead of a command list it gives a lookup loop: describe the task to cf cli search, check the request shape with cf schema or the options with --help, preview with --dry-run, then run.

Setup is npm install --global cf, which adds an identical cloudflare command in case another cf is on the PATH. cf auth login prints a link and a one-time code (--no-browser over SSH or in a container); unattended agents and CI use CLOUDFLARE_API_TOKEN.

A destructive command without --force in a non-interactive session prints Aborted. and still exits 0. --force can also be an API parameter: cf workers delete --force also deletes a Worker other Workers reference. --local works only for KV, some D1 commands and R2.

The skill also picks the CLI for a project: cloudflare.config.ts means cf; a Wrangler config means Wrangler for dev, build and deploy, with cf only for resource commands; neither means cf. Wrangler commands from product docs are translated, not run; live logs and single secrets, which cf lacks, go through npx wrangler. references/skill-feedback.md reports faults in the skill as a GitHub issue you approve first.

When to use it, and when not

Use it for DNS, zones, D1, KV, R2 and other API operations from the terminal, for new Workers projects and for projects with cloudflare.config.ts.

Not for dev, build or deploy in a Wrangler project.

How to run it

The install command below puts the skill folder in place; the CLI is installed separately, as shown, and needs Node.js 22.18 or later (not Bun). To call the skill yourself:

  • Claude Code: /cf-cli, or ask for the skill by name.
  • Codex: $cf-cli.
  • Cursor: type / in the Agent chat and pick cf-cli.

Example: /cf-cli Create a CNAME record for docs.example.com in my zone; show the dry run first.

The skill suggests a line for your user-level AGENTS.md or CLAUDE.md so agents prefer cf unless a Wrangler config exists.

How it differs from similar skills

Wrangler CLI covers the Workers lifecycle in Wrangler projects; this skill covers the whole API and new projects. Hugging Face Hub CLI embeds a generated list of hf commands in SKILL.md; with 2,900 commands, this one has the agent search instead. Sentry issues and events uses a similar sentry schema and sentry api pair, but only reads.

Install

Run one command — it saves SKILL.md and the files it uses into your agent's skills folder. Claude Code and Codex pick up a new skill without a restart; if it does not show up, restart the agent.

Claude Code — macOS, Linux
curl -fsSL --create-dirs -o ~/.claude/skills/cf-cli/SKILL.md https://anymodel.org/en/skills/cloudflare/cf-cli/SKILL.md -o ~/.claude/skills/cf-cli/LICENSE https://anymodel.org/en/skills/cloudflare/cf-cli/LICENSE -o ~/.claude/skills/cf-cli/references/skill-feedback.md https://anymodel.org/en/skills/cloudflare/cf-cli/references/skill-feedback.md
Codex, Cursor — macOS, Linux
curl -fsSL --create-dirs -o ~/.agents/skills/cf-cli/SKILL.md https://anymodel.org/en/skills/cloudflare/cf-cli/SKILL.md -o ~/.agents/skills/cf-cli/LICENSE https://anymodel.org/en/skills/cloudflare/cf-cli/LICENSE -o ~/.agents/skills/cf-cli/references/skill-feedback.md https://anymodel.org/en/skills/cloudflare/cf-cli/references/skill-feedback.md
Claude Code — Windows PowerShell
foreach ($f in 'SKILL.md','LICENSE','references/skill-feedback.md') { $p = Join-Path "$HOME\.claude\skills\cf-cli" $f; New-Item -ItemType Directory -Force (Split-Path $p) | Out-Null; Invoke-WebRequest "https://anymodel.org/en/skills/cloudflare/cf-cli/$f" -UseBasicParsing -OutFile $p }
Codex, Cursor — Windows PowerShell
foreach ($f in 'SKILL.md','LICENSE','references/skill-feedback.md') { $p = Join-Path "$HOME\.agents\skills\cf-cli" $f; New-Item -ItemType Directory -Force (Split-Path $p) | Out-Null; Invoke-WebRequest "https://anymodel.org/en/skills/cloudflare/cf-cli/$f" -UseBasicParsing -OutFile $p }

Any other agent: save the file into its skills folder, or give it the link to the file: https://anymodel.org/en/skills/cloudflare/cf-cli/SKILL.md

SKILL.md

Raw

Cloudflare CLI (cf)

cf is Cloudflare's command-line interface for the public Cloudflare API (2,900+ commands, most generated from API schemas) and for Workers projects. It is in beta: commands, configuration, and Build Output can change before the stable release.

Prefer retrieval over pre-trained knowledge. Describe the task to cf cli search, inspect the match with cf schema or --help, and preview with --dry-run before running a command you have not verified.

Docs index: https://developers.cloudflare.com/cf/llms.txt

Documentation

Topic Page
Install, sign in, credential and account selection https://developers.cloudflare.com/cf/get-started/index.md
First Worker with cf init, cf dev, cf deploy https://developers.cloudflare.com/cf/get-started/first-worker/index.md
Find, create, and delete resources from the terminal https://developers.cloudflare.com/cf/get-started/resources/index.md
cf for Wrangler users https://developers.cloudflare.com/cf/wrangler/index.md
Migrate a Wrangler project with cf migrate https://developers.cloudflare.com/cf/wrangler/migrate/index.md
Wrangler-to-cf command and config reference https://developers.cloudflare.com/cf/wrangler/reference/index.md
Develop, build, and deploy; modes and previews https://developers.cloudflare.com/cf/projects/index.md
Typed cloudflare.config.ts https://developers.cloudflare.com/cf/projects/cloudflare-config/index.md
Using cf with coding agents (safety rules) https://developers.cloudflare.com/cf/agents/index.md
CI, API tokens, unattended authentication https://developers.cloudflare.com/cf/ci/index.md
Environment variables https://developers.cloudflare.com/cf/environment-variables/index.md
Find slow or expensive D1 databases and queries with cf sql query https://developers.cloudflare.com/d1/observability/find-expensive-queries/index.md

FIRST: check installation and sign in

npm install --global cf   # installs two equivalent commands: `cf` and `cloudflare`
cf --version
cf auth whoami
  • Requires Node.js 22.18 or later. Bun is not supported.
  • If another tool named cf is on your PATH, use cloudflare instead — the package installs both commands.
  • cf auth login prints a link and a one-time code, then opens the browser. On a remote machine, over SSH, or in a container, add --no-browser so cf only prints the link; approve it from another device. Add --force to sign in again.
  • cf keeps its own credentials and does not reuse a Wrangler login.
  • Unattended agents and CI cannot complete cf auth login — set CLOUDFLARE_API_TOKEN instead (see https://developers.cloudflare.com/cf/ci/index.md).

To make agents prefer cf, add this to the user-level AGENTS.md, CLAUDE.md, or equivalent:

When interacting with Cloudflare, use the `cf` CLI unless the project has a
Wrangler configuration file.

Find commands

cf cli search "create a DNS record"   # local, no credentials; quote the whole task; up to 5 JSON matches
cf schema d1 create                   # API request shape: method, path, params, body fields
cf <command> --help                   # arguments and options (e.g. cf deploy --help)

cf cli search takes the task as one argument — unquoted words after the first are parsed as unknown commands. cf schema covers generated API commands; for command arguments and options use --help.

Read output

  • Results are JSON on standard output; status messages and errors go to standard error, so piping to jq or redirecting needs no extra flags.
  • Lists print one JSON array page; paging options differ per command — check --help.
  • A failed command exits non-zero and prints the error to standard error.

Run commands safely

  • Preview first. --dry-run prints the request as JSON without sending it, and needs no credentials.
  • Aborted deletes still exit 0. In a non-interactive session, a destructive command without --force prints Aborted. to standard error and exits 0 — a zero exit does not mean the resource was deleted.
  • Review --force before allowing it. On some commands --force is also an API parameter (for example, cf workers delete --force also deletes a Worker that other Workers reference).
  • --local is limited. It works for KV keys, D1 databases (cf d1 raw, cf d1 migrations list|apply), and R2 objects; other commands (including cf d1 query) return an error instead of calling the API.

Choose the CLI for this project

  • cloudflare.config.ts exists: use cf.
  • wrangler.jsonc, wrangler.json, or wrangler.toml exists, and no cloudflare.config.ts: keep Wrangler for dev, build, and deploy. Do not run cf dev, cf build, or cf deploy there — they can generate a new configuration that ignores the Wrangler file, or fail. cf resource commands (DNS, zones, storage) are fine alongside it. Run cf migrate only when the user asks, and follow Migrate a Wrangler project: preview with cf migrate --dry-run, run cf migrate, then resolve the TODO(@cloudflare) comments in the generated cloudflare.config.ts.
  • Neither exists: use cf, as Use cf with coding agents recommends. Do not install Wrangler or create a Wrangler configuration file.

Product documentation often shows only Wrangler commands and wrangler.jsonc configuration. In a cf project, treat those as tasks to translate, not commands to run, and do not add a Wrangler configuration file. Use one CLI per project; the only exceptions are the commands under Commands cf does not support yet. For Wrangler equivalents, use the Wrangler to cf reference or cf cli search. Do not translate Wrangler commands from memory. For the full comparison, refer to cf for Wrangler users.

Commands cf does not support yet

cf does not support every Wrangler command yet. Run these with npx wrangler instead of adding Wrangler to the project. Wrangler does not read cloudflare.config.ts, so pass the Worker name:

  • Live logs: npx wrangler tail <WORKER_NAME>.
  • A single secret: npx wrangler secret put <SECRET_NAME> --name <WORKER_NAME>. Or upload secrets with a new Worker version by passing --secrets-file <PATH> to cf deploy or cf workers versions create.

For the current list, refer to "Commands not yet supported" in the Wrangler to cf reference.

Skill feedback

If the user reports that this skill gave incorrect or outdated guidance, or is missing information, read references/skill-feedback.md.

Connect through AnyModel API

This skill runs on whatever model your agent uses. Point Claude Code, Codex or Cursor at AnyModel to run it on Claude, GPT, Gemini and more with one key — pay per token, no subscription.

More from the pack Cloudflare skills

Cloudflare Workers best practices

Checks Cloudflare Workers code and config against production rules: current compatibility dates, logs and traces, streaming, waitUntil, bindings over REST calls, Web Crypto, generated Env types. Use when writing, reviewing or shipping a Worker.

Wrangler CLI

Guides Wrangler commands and config in projects with wrangler.jsonc or wrangler.toml, checking installed help and current docs and guarding secrets, environments and Previews. Use when deploying, configuring or debugging a Worker with Wrangler.

Cloudflare Durable Objects

Builds and reviews Cloudflare Durable Objects: one object per entity, SQLite storage and migrations, RPC methods, alarms, WebSockets and Vitest tests, with links to current docs. Use when a Workers app needs shared state or coordination.

Cloudflare Agents SDK

Builds and reviews apps on Cloudflare's Agents SDK: stateful agents on Durable Objects, callable RPC, scheduling, queues, Workflows, AIChatAgent chat, MCP and email. Use when building an AI agent or chat app on Workers.

Web frameworks on Cloudflare Workers

Sets up Astro, SvelteKit, React Router, TanStack Start, Nuxt, Hono and SPA apps on Cloudflare Workers, choosing C3, Wrangler or cf by the project's state. Use when creating, migrating or deploying a framework app on Workers.

Local Workers development

Runs and debugs Workers locally with cf dev, wrangler dev or the Vite plugin, using Local Explorer to read and seed binding data and query traces. Use when a local request fails or you need to see what a binding returned.

Web performance audit

Audits a web page's loading performance through Chrome DevTools MCP: a trace, Core Web Vitals against thresholds, network, an accessibility snapshot and bundle config, with ranked fixes. Use when a page is slow or its Lighthouse score is low.