Agents
Give your agent the same CLI you use
Eleven skills ship in the npm package: deploying, reading logs, managing domains, members and billing. They drive the CLI and hold no credentials.
$ npm install -g @tensel/cli$ tensel skills installInstalled 11 skills into ~/.claude/skills. $ tensel loginOpening your browser to approve this device…✓ Signed in.No credentials in the skill
A skill drives the CLI and never calls the API itself. “tensel login” stays the only path a credential takes, and an API key is revocable at any time.
The limits are written down
Every skill states what the platform does not do, so the agent tells you instead of improvising a workaround.
“--json” on every command
Every read command emits documented JSON. The JSON shape is the contract; the prose output is for humans and may change.
Secrets stay secret
Secrets are write-only: set once, injected at deploy. Nothing can read a value back, skills included.
Your sites answer agents too
Ship a Markdown copy of a page beside it and your site serves that to clients asking for Markdown, and the HTML to everyone else. It works on every plan and there is nothing to switch on.
The eleven skills in the box
Installed into your agent's skills folder by “tensel skills install”.
tensel-getting-startedSign in and find your way around an account.
tensel-deployDeploy from a folder, a repo or a commit; cancel and roll back.
tensel-githubConnect a repository so pushes deploy, and fix bindings that do not.
tensel-environmentsCreate and manage projects, applications and environments.
tensel-build-settingsUnderstand and change how an application builds.
tensel-config-secretsSet variables and secrets, with build or runtime scope.
tensel-domainsAttach a custom domain and diagnose one that will not verify.
tensel-logsRead build and application logs to find out what happened.
tensel-membersInvite people, change roles, revoke access.
tensel-billingPlans, usage, invoices and spend limits, explained in place.
tensel-troubleshootStart here when something is wrong and the cause is unclear.
This site also publishes /llms.txt: the platform's essential facts, in the format agents read before they act.