flompt is an open-source visual prompt builder designed to transform unstructured text into highly structured, optimized prompts. Created by Nyrok, the tool is built on the philosophy that prompt structure is the most significant factor in improving AI output quality. By decomposing a single block of text into discrete, functional components, flompt allows users to manage the complexity of prompt engineering through a visual interface. The platform is free to use, requires no account, and is available as a web application, a browser extension, and a Model Context Protocol (MCP) server.

The tool functions by breaking any input prompt into 16 distinct block types, such as roles, objectives, and constraints. Once these components are edited and refined, flompt recompiles them into a Claude-optimized XML format. This structured approach helps in providing clear grounding and context for large language models, particularly those that perform better with XML-tagged instructions. The tool is designed to be a bridge between a user's initial idea and a machine-ready instruction set that minimizes hallucinations and improves task adherence.

Some of the key features are:

• Visual Block Builder: Breaks down complex prompts into 16 different typed blocks including Role, Objective, Constraints, and Examples for easier management.
• Claude-Optimized XML Compilation: Automatically formats the final output into structured XML tags like , , and to maximize performance on Claude and other models.
• Browser Extension Integration: Adds a toolbar button to ChatGPT, Claude, and Gemini interfaces to allow prompt building and injection directly within the chat window.
• Model Context Protocol (MCP) Server: Provides native integration with Claude Code, allowing AI agents to call tools like decompose_prompt and compile_prompt during autonomous tasks.
• Prompt Audit and Scoring: Scans prompts for missing elements like roles or examples and provides a score from 0 to 100 to help users identify and fix weaknesses.
• Template Library: Includes over 100 pre-configured templates across categories like code, writing, marketing, design, and sales to eliminate blank-page syndrome.
• Context Memory: Allows users to store persistent facts about their technology stack, brand tone, or specific rules that are automatically loaded into every new prompt.
• Make.com Webhook Integration: Enables users to send assembled prompts, formats, and metadata directly to Make.com automation workflows with a single click.
• Version History and Diff View: Saves prompt iterations with labels and allows side-by-side comparisons to see exactly what lines were added, removed, or moved.

To use flompt, a user typically starts by pasting an existing prompt or selecting a template from the library. The interface immediately decomposes the text into various blocks which can be reordered, edited, or deleted. Users can then utilize the Audit tool to check for structural gaps. Once the prompt is refined, clicking the compile button generates the XML version, which can then be copied manually, sent to a webhook, or injected directly into a supported AI chat interface via the browser extension. For developer workflows, the tool can be run as a local MCP server or self-hosted via Docker.

Some common use cases include:

• Developer Documentation: Using the README Generator or API Docs templates to create structured technical documentation from raw code snippets.
• Marketing Content Generation: Developing landing page copy or email sequences by injecting brand-specific context memory and style constraints.
• AI Agent Instruction: Utilizing the MCP server within an agentic workflow to programmatically refine prompts for sub-tasks.
• Legal and Professional Writing: Structuring complex constraints and guardrails for sensitive professional emails or contract summaries.
• Code Quality Assurance: Leveraging the Code Review and Unit Test Writer templates to provide structured feedback on software projects.