Laravel, Node.js, Python, Git and Logs Are Now in the Panelica API (and Your AI Can Use Them)

Yes: Panelica's External API includes 21 Laravel endpoints, 18 for Node.js, 19 for Python, 43 for Git and deploy, and 26 for logs, and every one of them is exposed as an MCP tool an AI assistant can call. This means an AI connected to your panel is not limited to classic shared-hosting tasks — it can deploy, restart, inspect and debug real application runtimes the same way a developer would.

Why this matters more than it sounds like

Most of the conversation about AI and hosting panels focuses on classic shared-hosting tasks — create an account, provision a domain, install WordPress. Those are real and useful, and we cover them in Create a WordPress Site by Talking to an AI. But a growing share of what actually runs on a Panelica server is application code, not PHP-and-MySQL sites in the traditional sense — Laravel APIs, Node.js services, Python applications, all deployed from Git. An AI assistant that can only manage domains and databases is not actually useful to the developer running that stack day to day. One that can also deploy a branch, restart a crashed worker, or pull an application's error log is.

Laravel: 21 endpoints

The Laravel-specific portion of the API covers application lifecycle beyond generic PHP hosting: environment configuration, queue worker management, scheduled task status, and artisan-level operations exposed as structured API calls rather than requiring shell access. An AI assistant with the right scope can, for example, check whether a queue worker has stalled and restart it, or verify a scheduled command's last run time — the kind of routine Laravel operations work that would otherwise mean SSH and a memorized list of artisan commands.

Node.js: 18 endpoints

Node.js applications get their own endpoint family covering process management, runtime version selection, environment variables and restart control. This is what lets an AI-driven request like "restart the Node app after the latest deploy" resolve into an actual process restart through the panel's process manager, rather than a shell command run blind.

Python: 19 endpoints

Python Application Hosting, covered in full in Announcing Python Application Hosting, ships with its own 19-endpoint API surface for deploying and operating Python services — virtual environment and dependency handling, process supervision, and application-level logs. Every one of those 19 endpoints is available as an MCP tool, so the walkthroughs in that announcement are things an AI assistant connected over MCP can carry out directly, not just things a human can do through the dashboard.

Git and deploy: 43 endpoints, the largest single family

Git and deploy is the single biggest endpoint family in the API at 43 endpoints — repository connections, branch management, deploy hooks, and deployment history. This is the family that makes a request like "deploy the latest commit on main to production" a real, traceable action: the AI calls the same deploy tooling a CI pipeline or a developer's manual deploy would use, and the deployment shows up in the panel's own deploy history afterward, not just in a chat transcript.

Logs: 26 endpoints

Every one of the runtimes above, plus the classic web and mail stack, generates logs, and 26 endpoints exist specifically to read them — service logs, application logs, error logs, filtered by time range and severity. This is what turns "why is the app returning 500s" from a request that requires SSH and grep into one the AI can answer directly, by pulling the actual recent error log through the API and reading it.

How this composes in a real session

These families are not used in isolation. A realistic debugging request touches several of them in sequence:

You: The Laravel API is returning 500s since the last deploy. What changed and can you fix it?

tool: git.deployments.list -> most recent deploy, commit hash, timestamp
tool: logs.application.tail -> recent Laravel error log entries
tool: laravel.queue.status -> queue worker state
[assistant identifies a missing environment variable introduced by the new commit]
tool: laravel.env.set -> adds the missing variable
tool: laravel.queue.restart -> restarts the affected worker
tool: logs.application.tail -> confirms errors have stopped

Each of those is a real, separately scoped API call — the investigation and the fix both happen through the panel's actual tooling, not a guess.

This is why the tool count keeps growing

These endpoint families exist because Panelica treats application hosting as core functionality, not an add-on bolted next to classic shared hosting — and because the MCP tool catalogue is generated from the live API rather than hand-maintained, every one of these families is available to AI clients automatically. See How We Keep 404 AI Tools in Sync with a Live API for how that generation process works.