Prompt Babbler

💬 Prompt Babbler is a speech-to-prompt web application that captures stream-of-consciousness speech, transcribes it using Microsoft Foundry, and generates structured prompts for target systems like GitHub Copilot. It can run locally with a Cosmos DB emulator or be deployed to Azure with a fully managed Cosmos DB instance and Microsoft Foundry resources. The app is built with a React frontend and a .NET backend, orchestrated by .NET Aspire for seamless local and cloud development.

What It Does

Prompt Babbler turns rough speech into polished, structured prompts ready to use with AI tools. The typical workflow is:

1. Record — Click record and speak your thoughts out loud, stream-of-consciousness style.
2. Transcribe — Azure AI Speech Service converts your audio to text in real time.
3. Generate — The app sends your transcription through a configurable prompt template and calls a Foundry Models model to produce a structured, ready-to-use prompt.
4. Use — Copy the generated prompt into GitHub Copilot, an AI assistant, an image generator, or any other AI tool.

Key Features

• Real-time speech transcription via Azure AI Speech Service — see your words appear as you speak.
• Prompt templates — Create and manage reusable templates that shape how your transcription is turned into a prompt. Templates support structured instructions, output format, guardrails, examples, and tags.
• Prompt history — Every generated prompt is saved alongside its source transcription (called a babble) so you can review and reuse past outputs.
• Multi-target support — Included built-in templates for GitHub Copilot, general AI assistants, and image generators.
• Single-user and multi-user modes — Run privately without authentication, or enable Microsoft Entra ID for multi-user access.
• Fully cloud-native — Deploys to Azure Static Web Apps + Azure Container Apps with Cosmos DB and Microsoft Foundry, all provisioned via Bicep.
• MCP server — Exposes babbles, templates, and prompt generation to GitHub Copilot, Claude, and any MCP-compatible AI client via the Model Context Protocol.

Quick Start

For full setup instructions including prerequisite installation, see docs/quickstart-local.md. To deploy to Azure, see docs/quickstart-azure.md.

Prerequisites to run Locally

• An Azure Account with Contributor access to allow Aspire to provision Microsoft Foundry resources for the app
• Aspire CLI to orchestrate the components
• Azure CLI to authenticate and allow Aspire to provision Microsoft Foundry resources in your subscription
• Install Docker Desktop to host Cosmos DB emulator (Windows/Linux/Mac)

Run Locally

git clone https://github.com/PlagueHO/prompt-babbler.git
cd prompt-babbler

# Sign in to Azure (one-time — Aspire provisions cloud AI resources)
az login --tenant <your-tenant-id>

# Start everything via Aspire
aspire run

Aspire handles all dependency installation, builds, and orchestration automatically. On first run it provisions Microsoft Foundry resources and starts a Cosmos DB emulator in Docker — this takes several minutes. Subsequent runs start quickly.

---

MCP Server

Prompt Babbler includes an MCP (Model Context Protocol) server that exposes your babbles, prompt templates, and prompt generation to any MCP-compatible AI client. Connect GitHub Copilot, Claude, or another agent and let it search your voice notes or generate prompts on your behalf — without leaving the chat interface.

The MCP endpoint when running locally is http://localhost:5242. See docs/mcp-server.md for the full tool reference, resource catalog, and authentication options.

VS Code (GitHub Copilot Agent Mode)

Create .vscode/mcp.json in your workspace:

{
  "servers": {
    "prompt-babbler": {
      "type": "http",
      "url": "http://localhost:5242"
    }
  }
}

Open GitHub Copilot Chat, switch to Agent Mode, and prompt-babbler appears in the tools list.

Claude Code

claude mcp add --transport http prompt-babbler http://localhost:5242

GitHub Copilot CLI

gh copilot mcp add prompt-babbler --transport http http://localhost:5242

---

Documentation

Document	Description
Architecture	Tech stack, project structure, API endpoints, data model, infrastructure
Local Development	Run locally with .NET Aspire
Deploy to Azure	Deploy to Azure with Azure Developer CLI
API Reference	Full API reference with request/response schemas
MCP Server	MCP tools, resources, prompts, and client configuration
CI/CD Setup	GitHub Actions pipeline configuration
Infrastructure	Azure Bicep infrastructure details

License

MIT

Contributing

Contributions are welcome. Please open an issue to discuss proposed changes before submitting a pull request.

See CONTRIBUTING.md, CODE_OF_CONDUCT.md, SECURITY.md, and SUPPORT.md for the project community guidelines.