Whisk CLI

The command line your agent uses: sign in, check, deploy, read logs.

Plans: Every plan, including the free app

What you get

Sign-in approved by a person in the browser
Check, deploy, roll back and read logs
JSON output and fixed exit codes for agents
Every error with a code and a fix
Self-update with signature checks
Open source under the MIT licence

How it works

whisk login prints a link and a code, and a person approves it in the browser. The token stays on that computer, in the system keychain where there is one. After that, the agent can create apps, check them, deploy them and look after them.

Every command takes --json and ends with a documented exit code, including one that means a person must act. Errors carry a stable code and a fix, so the agent knows what to do next. whisk doctor checks an app before it is pushed.

whisk update downloads new versions and checks their checksum and signature before installing. The CLI is written in Go and published under the MIT licence at github.com/whisk-hosting/cli, along with the agent skill.

Example

An HVAC contractor's coding agent changes the service scheduling app and deploys it. The health check fails, so the CLI returns HEALTH_CHECK_FAILED with a log excerpt. The previous version stays live while the agent fixes the port and deploys again.

For your coding agent

curl -fsSL https://whisk.run/install.sh | sh
whisk login
whisk doctor
whisk deploy
whisk logs -f

Every command and flag is in the CLI reference.

All features