# Agentic Coding Toolset

The Agentic Coding Toolset brings Form.io into an AI coding environment. The Agentic Coding Toolset is what an AI agent uses while writing code. It can plan data models, generate forms, configure access controls, and scaffold front ends on the Form.io platform.

## What's in the toolset?

The toolset is delivered as a single Claude Code plugin (@formio/ai) that bundles three components. It includes the following components:

* **Claude Code plugin (@formio/ai)** - Registers the MCP server and skills library with Claude Code. Can be installed with a single command.
* **MCP Server (@formio/mcp)** - Exposes Form.io operations (form\_, role\_, action\_, project\_) as MCP tools the agent can call. Works with any MCP-aware client such as Claude, Visual Studio Copilot, or Bolt.
* **Skills Library** - Enables covering app orchestration, resource planning, framework scaffolding, form authoring, action configuration, and the full Form.io REST surface.

## How the toolset works

When an agent receives instructions on what to build, it reviews the relevant skills to understand Form.io's primitives, then calls MCP tools to make changes in the Form.io project. The available skills are used to shape the pattern of the solution, while the MCP tools give it access to the functions necessary to accomplish that.

Every app the agent builds inherits the model, the governance, RBAC, and the audit trail by default. Approval gates run at every write — the agent doesn't make changes without your review.

### What can the toolset do?

The toolset handles solutions of varying scale and complexity — building from scratch, extending an existing app, modifying access controls, planning a data model, and operational queries against a live project. 

Three examples:

#### Build a complete app from one prompt:

*"Build me a CRM where sales reps only see accounts owned by the teams they belong to."*

The agent plans the data model, imports it into Form.io, and scaffolds an Angular front-end wired to the project. At the end, it delivers a running application.

#### Tighten access controls on an existing resource.

*"Currently anyone authenticated can read every Account. Lock it down so reps only see Accounts owned by their Team."*

The agent reads the resource, modifies the access rules, and ensures the supporting role assignments are in place. The diff is provided for review before anything ships.

#### Plan a data model from plain language.

*"Plan the resource structure for a multi-tenant booking system where customers book services from providers, providers belong to one or more locations, and admins manage everything."*

The agent runs a structured interview, produces an ER diagram and access matrix for review, and emits a project template ready to import.

#### Additional Examples

See more examples in the [Form.io Agentic Coding Toolkit GitHub repo](https://github.com/formio/ai).

## Getting started

The toolset installs as a Claude Code plugin:

1. From inside Claude Code, run

```
/plugin install formio-ai@formio
```

2. A prompt will request **FORMIO\_BASE\_URL** and **FORMIO\_PROJECT\_URL.**
3. When installation is complete, it may be necessary to restart the terminal session or desktop application.

The agent can now respond to natural langauge requests, documents, or images to begin working on a project.

For full setup instructions, including the standalone MCP server for use with Claude Desktop, VS Copilot, or other MCP-aware clients, see the [Form.io Agentic Coding Toolkit GitHub repo](https://github.com/formio/ai).


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://help.form.io/ai-and-form.io/agentic-coding-toolset.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.