Agent skill

What you get

• One address for any coding agent
• The rules every Whisk app follows
• Every whisk.yaml setting, with examples
• Every error code, with its fix
• Checks the agent runs before it deploys
• Open source under the MIT licence

How it works

Give your coding agent the address whisk.run/skill. It reads one document that explains how a Whisk app works. It covers the rules, the whisk.yaml file, sign-in, the database, workflows, webhooks, storage and email.

The document links to the files an agent checks its work against: the whisk.yaml schema, the error catalogue and the whisk doctor rules. The same document is printed by whisk skill and published as the getting started docs, so people and agents read the same thing.

The skill, the whisk.yaml format, the templates and the CLI are open source under the MIT licence, at github.com/whisk-hosting/cli. It works with Claude Code, Codex, Cursor, Gemini CLI and other agents that can edit files and run commands. See the Claude Code guide.

Example

A steel fabricator's office manager asks a coding agent for a tool that tracks jobs through the workshop. After reading the skill, the agent sets up sign-in, the database and the deploy, then lists the secrets a person needs to set.

For your coding agent

whisk skill
whisk init --template ts
whisk doctor
whisk deploy

Or tell the agent to read whisk.run/skill before it starts. The commands are in the CLI reference.