Skip to content

Modules · Veva.Modules.MCP

Ask the site. Agents can too.

Turn on the AI assistant and editors ask for pages, documents and images in plain words, with the permissions they already have. The same tools are an MCP server at /mcp, so coding agents build against the running site.

dotnet add package Veva.Modules.AI.OpenAI

The AI assistant

Editors ask. The assistant does the clicking.

Veva.Modules.AI.OpenAI provides the model. The assistant's loop and tools come with Veva.

In the admin

Turn it on per website, and an AI assistant button in the admin opens a chat about that website.

The editor's own tools

It adds pages and components, writes and translates documents, files assets and edits menus.

Your permissions, no more

Every tool call is checked against the permissions of the person asking, as the admin checks them.

It asks first

Creating a model, deleting data, changing many documents or a model's fields waits for a yes.

Files and images

Attach images, PDFs or text files. With an image provider set, it generates images into the library.

Your provider, your prompt

Pick the AI provider per website, such as OpenAI, and steer every conversation with a system prompt.

The MCP server

One server, tools from every module.

Veva.Modules.MCP owns /mcp, and each module adds its tools to it. The Pages, Documents and Assets tools come with Veva.Web.Essentials.

Pages and content

Create and move pages, add components, import a content tree, render a page and publish it.

Documents

Design models from templates, search, write and translate documents, update many at once.

Assets

Upload files or fetch them from a URL, search the library, set up folders, generate images.

Menus, routes and tags

Hand-built menus, routing rules with a route tester, and tags in categories.

Answers about Veva

get_type_info describes any public Veva type, and list_components reads each component's hint.

Mistakes you can undo

Deleted pages, content and documents can be restored. Page content is saved as a draft.

How an agent connects

From a key to a checked page.

  1. 01

    Get a key

    An api key, or in development the agent account BuildVeva() sets up.

    Veva:DevAuth
  2. 02

    Connect

    Point the agent's MCP client at /mcp, with the key as a bearer token.

    /mcp
  3. 03

    Look around

    The agent reads the site's templates, pages and components first.

    get_site_overview
  4. 04

    Build

    It adds pages, components, documents and assets, as an editor would.

    add_content
  5. 05

    Check

    It reads the HTML a page serves and the issues the designer would show.

    render_page

For developers

Connect Claude Code, or any MCP client.

Add the server to the client's settings. The X-Veva-Tool-Confirmation header says the client asks its user before each tool call, so the tools that need a yes can run. Setting it is the developer's decision.

Extension points

API

Extend the assistant and the server from a module:

  • IAiChatProvider: an AI provider for the assistant
  • IAiImageProvider: image generation for the assistant
  • IMcpToolPermissions: who may run your module's tools
  • AgentHint: tells agents what your component is for
.mcp.json
{
  "mcpServers": {
    "veva": {
      "type": "http",
      "url": "https://localhost:5001/mcp",
      "headers": {
        "Authorization": "Bearer ${VEVA_AGENT_KEY}",
        "X-Veva-Tool-Confirmation": "client"
      }
    }
  }
}

Modules

What it is made of.

The ones marked Built in come with Veva.Web.Essentials. Add the others as packages, restart, and set them up in the site's settings.

Module library

MCP server

Veva.Modules.MCP

The MCP server at /mcp and the admin's AI assistant. Each module adds its tools: pages, documents, assets, tags, menus and routes.

Built in
dotnet add package Veva.Modules.MCP

Content import

Veva.Modules.ContentImport

Imports Word, PDF and image files as page content: upload, review the proposed components, then insert them.

Add-on
dotnet add package Veva.Modules.ContentImport

OpenAI

Veva.Modules.AI.OpenAI

OpenAI chat and image generation for the Veva AI assistant.

Add-on
dotnet add package Veva.Modules.AI.OpenAI

More of Veva

Explore the other features.

Each one is made of modules. Reference a package, restart, and its components are in the editor.

All features

Page builder

Compose pages from components, then preview and publish

Open

Components and data

Components that show, components that fetch, and the binding between them

Open

Documents

Structured content you list, filter and serve anywhere

Open

Forms

Document forms that save, generic forms that filter or post

Open

PDFs and signing

Form answers made into a PDF, downloaded or sent to be signed

Open

Assets

One media library, with images sized on request

Open

Search

Crawl, index and search the whole site

Open

Veva core

BuildVeva(), modules, components, themes and hooks

Open

Tech stack

Plain .NET Blazor: three render modes, standard components

Open

Integrations

Data sources, sign-in, mail, analytics and media

Open

Module library

Every module: what comes with Essentials, and the add-ons

Open

Type to search the docs, components and API.

↑↓ to move ↵ to open Esc to close Search by Veva