Agentic Forms: A Dual-Mode Form Builder for Humans and AI Agents

Research context and background

Agentic Forms is a Next.js application built with SQLite and Prisma that enables the creation of forms accessible to both humans and AI agents. The core innovation is a dual-mode architecture: published forms present a standard web interface for human users while simultaneously exposing a machine-readable schema embedded in the page. This allows AI agents to discover, understand, and submit forms via a plain JSON POST request.

Benefits

The main advantage of Agentic Forms is its ability to bridge the gap between human interaction and automated processing. It provides a standard web form visible to users while also exposing a machine-readable schema embedded in the page. This allows AI agents to discover the form, understand field constraints, and submit data programmatically. The system includes a demo agent that can discover embedded forms, answer them based on a profile, and submit the data. It features self-correction logic that retries once on validation errors. The platform supports both API key usage for advanced mapping and deterministic keyword matching for basic operation without external dependencies.

Use Cases

This tool is ideal for organizations that want to automate form submissions while maintaining a user-friendly interface for humans. It can be used to create scholarship applications, job applications, or any form-based data collection process that benefits from AI automation. The system includes a Model Context Protocol server that allows Node scripts to search for forms, fetch schemas, and submit answers. It integrates with AI assistants like Claude Desktop through configuration files. The platform is suitable for developers who want to build forms that can be discovered and filled out by AI agents without changing the user experience for human visitors.

Pricing

No pricing information is available for this open-source project.

Vibes

No public reviews or testimonials are available for this project.

Additional Information

The project is built using Next.js with the App Router, SQLite for the database, and Prisma as the ORM. It runs on Node.js version 20 or higher. The system uses newline-delimited JSON-RPC 2.0 for the MCP server. Key environment variables include FORMYURL which defines where the MCP server sends requests, and APPURL which defines the public origin embedded in URLs. The project includes comprehensive testing with unit tests, integration tests, and end-to-end tests. Security considerations include no authentication for search and submission endpoints, and no request timeout for MCP calls. The data model consists of Form, Field, and Submission entities with an optional Agent Context block for form-wide guidance.