Skip to content

CONCEPT Cited by 1 source

CLI convention enforcement

CLI convention enforcement means baking naming and flag conventions into the schema layer of a CLI's code-gen pipeline so that inconsistency becomes a compile-time impossibility, not a review-time judgment call (Source: sources/2026-04-13-cloudflare-building-a-cli-for-all-of-cloudflare).

Why it matters

Large engineering orgs with many products accumulate inconsistencies: one team uses info, another uses get; one uses --skip-confirmations, another --force; one supports --json, another --format. Manual code review catches some of this but, as Cloudflare says, becomes "Swiss cheese" at hundreds of reviewers across ~100 products.

Agents pay the cost. An agent that learned one product's CLI will naturally try the same pattern on another product and call a command that doesn't exist — or worse, call a command that exists with different semantics and silently corrupt state.

Cloudflare's rules (as of 2026-04-13)

  • Always get, never info.
  • Always --force, never --skip-confirmations.
  • Always --json, never --format, and --json supported across every command with structured output.
  • Local / remote behavior explicitly signaled in every command's output.

These rules are enforced at the schema layer — the TypeScript-based internal schema that generates the cf CLI, Workers bindings, the SDK, the Terraform provider, and the MCP server. A product's command definition cannot be committed if it uses a disallowed verb or flag. See concepts/unified-interface-schema.

Also applies across interfaces

Enforcement isn't just a CLI concern. The same naming rules must propagate to:

  • SDK methods (client.foo.get() not client.foo.info()).
  • Terraform resource arguments.
  • MCP tool names (agents that read one tool set shouldn't trip on a differently-named sibling).
  • Configuration schema (wrangler.jsonc fields).

Because every surface is generated from one schema, the convention rules only need to live in one place.

Comparison to manual review

Manual review produces probabilistic consistency; schema-layer enforcement produces deterministic consistency. The author frames it as:

"In a large engineering org of hundreds or thousands of people, and with many products, manually enforcing consistency through reviews is Swiss cheese."

Seen in

Last updated · 200 distilled / 1,178 read