io.github.swantronPublished by GitHub account swantronv1.0.0Updated Oct 24, 2026
Generate a recipe from ingredients on hand, with optional dietary restrictions. Free.
A remote MCP server that turns ingredients on hand into a complete recipe, with optional dietary restrictions.
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.
Read https://sourceweft.com/mcp/io.github.swantron%2Fchomptron/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.Add this to your client's mcpServers config.
{
"mcpServers": {
"chomptron": {
"type": "http",
"url": "https://chomptron.com/mcp"
}
}
}AI-powered recipe generator that transforms ingredients into delicious recipes using Google Gemini AI. Deployed on Google Cloud Run at chomptron.com.
Enter ingredients you have in your kitchen, and Chomptron generates creative, practical recipes complete with measurements, instructions, cooking time, and serving sizes.
Visit http://localhost:8080
Get API key: https://makersuite.google.com/app/apikey
GEMINI_API_KEY (required) - Your Google Gemini API keyGEMINI_MODEL (optional) - Model to use, defaults to gemini-3.1-flash-litePORT (optional) - Server port, defaults to 8080The Gemini model can be configured via the GEMINI_MODEL environment variable:
Recommended Models (as of December 2025):
gemini-3.1-flash-lite (default) - Best free tier limits: 15 RPM, 1,000 RPDgemini-2.5-flash - 10 RPM, 250 RPDgemini-2.0-flash - 10 RPM, 200 RPD (⚠️ unstable quota, often shows limit: 0)gemini-1.5-flash - Legacy model, may have better limits than 2.0December 2025 Quota Shift:
Google overhauled free tier quotas in December 2025:
gemini-2.0-flash was removed from fully unauthenticated free tierlimit: 0 errors for newer models without billing enabledFixing "Limit: 0" Errors:
If you're seeing quota errors with limit: 0:
gemini-2.5-flash-lite - Best free tier model currently available/api/usage endpoint or https://ai.dev/usageFree Tier Limits (as of Dec 2025):
Tests validate:
styles.css and app.js linking/health, /ready) and recipe generationGET /health - Liveness check, returns service status
GET /ready - Readiness check, verifies AI configuration and shows current model
GET /api/usage - Usage statistics (total requests, quota errors, model info)
POST /api/generate-recipe - Main recipe generation endpoint
POST /mcp — MCP Streamable HTTP. Tool generate_recipe (ingredients + optional dietary flags). Unauthenticated. Daily cap MCP_DAILY_RECIPE_LIMIT (default 20), persisted to MCP_BUDGET_FILE so a process restart on the same instance does not reset it.
Chomptron is built as a serverless application on Google Cloud Run for cost efficiency and automatic scaling.
Tech Stack:
Why Serverless?
Performance Optimizations:
Health checks:
View logs:
Console dashboards:
Recipes are stored in browser localStorage with the following structure:
The application uses in-memory caching on the server side:
Optimized for search engines and social sharing with meta tags, Open Graph, Twitter Cards, structured data (JSON-LD), sitemap, robots.txt, and PWA support.
This repo includes a Buildkite pipeline at .buildkite/pipeline.yml that runs on a self-hosted GCP agent provisioned by buildkite-gcp-agent.
The key difference from the GitHub Actions workflow, which runs format → lint → test sequentially in a single job: these three steps have no dependencies on each other and run simultaneously. On a pool of agents this distributes the work; even on a single agent the explicit dependency graph documents intent and makes the pipeline trivially scalable.
The deploy stage is intentionally omitted — Cloud Run deployments are handled by the existing GHA workflow which holds the production GCP credentials. CI (fast feedback) and CD (production access) are kept in separate systems as a security boundary.
Connect the repo via Buildkite → New Pipeline → point at github.com/swantron/chomptron. Buildkite reads .buildkite/pipeline.yml automatically on each push.
All steps run on the gcp queue:
This is a personal project, but suggestions and improvements are welcome!
MIT
Source: README.md at commit d433aac
Generate a recipe
Given ingredients on hand (and optional dietary restrictions), generate a complete recipe: name, servings, timing, ingredient list, and steps.
ingredientsrequireddietaryPreferences100% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
npm installexport GEMINI_API_KEY="your-api-key-here"export GEMINI_MODEL="gemini-3.1-flash-lite" # Optional, defaults to gemini-3.1-flash-litenpm startnpm test{ "ingredients": "chicken, tomatoes, garlic", "dietaryPreferences": { "vegan": false, "vegetarian": false, "glutenFree": false, "dairyFree": false, "nutFree": false, "shellfishFree": false, "eggFree": false, "soyFree": false }}{ "success": true, "recipe": "**Recipe Name:** ...", "cached": false}curl https://chomptron.com/healthcurl https://chomptron.com/readycurl https://chomptron.com/api/usage # View usage statistics# Recent logsgcloud run logs read chomptron --region us-central1 --limit 50
# Live streamgcloud run logs tail chomptron --region us-central1{ "id": "timestamp", "ingredients": "chicken, tomatoes, garlic", "recipe": "Full recipe text...", "recipeName": "Extracted recipe name", "timestamp": "2025-01-01T00:00:00.000Z", "favorite": false, "rating": 0, "notes": "", "servingSize": 4, "dietaryPreferences": { "vegan": false, "vegetarian": false, "glutenFree": false, "dairyFree": false, "nutFree": false, "shellfishFree": false, "eggFree": false, "soyFree": false }}Push / PR └── Three steps dispatched in parallel: ├── :prettier: Format check (npm run format -- --check) ├── :eslint: Lint (npm run lint) └── :nodejs: Tests (npm test) └── Annotate — consolidated pass/fail surfaced in the Buildkite UIagents: queue: gcp