Who is it for?
Developers and AI-assisted builders with a working project on their machine but no repository, no Dockerfile experience, and no server to manage.
How does it work?
cdnctl init detects the project (and any local AI agents) and writes cdnctl.yaml plus a Dockerfile template; cdnctl check scans for production risks locally; cdnctl deploy uploads the folder, builds it in an isolated sandbox, and returns a live URL with SSL.
Included services
Dockerfile template, local pre-flight checks, isolated image build, subdomain with Auto SSL, status and logs — plus deploy-only tokens and a built-in MCP server so AI agents can ship without panel credentials.
Example scenario
A Node.js app written with an AI assistant goes from folder to live URL in one command; the agent then ships later updates itself using a deploy-only token.
Logs and cache from the same terminal
cdnctl is not only for deploys. cdnctl logs pull fetches a day of your edge access logs, cdnctl logs grep searches a time window on your own machine by IP, status code, path or country, cdnctl waf logs shows what the firewall blocked and cdnctl purge clears cached paths. The same token and the same saved account work for all of them, so an incident can be followed from the first alert to the fix without opening a browser.
Example scenarios
Go live without git, registry, or CI setup.
Let the agent that wrote the code deploy it, safely.
Catch localhost binds, leaked secrets, and image mistakes before upload.
Frequently asked questions
Do I need GitHub or Docker knowledge?
No. A project folder is enough: init writes the Dockerfile template, the platform builds the image in an isolated sandbox, and the result is a live HTTPS URL.
Can my AI agent deploy without my panel password?
Yes. A deploy-only token is shown once, works for a single account, and covers only the deploy surface — no delete and no billing. cdnctl mcp exposes the same commands over MCP.