Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@ReceiptCal show me what I ate yesterday and suggest a quick dinner recipe"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Receipts do the logging. Connect your own ReceiptCal account to an AI assistant to inspect food logs, request recipe ideas, and prepare confirmed pantry, planning and support actions.
This is the MIT-licensed developer integration, not the ReceiptCal iOS/Android apps or hosted backend. It is distributed directly on GitHub and is not an approved listing in the OpenAI plugin directory. A GitHub release does not change host eligibility, approval requirements, service terms or account permissions. Hosted use requires a ReceiptCal account and a client that supports OAuth and Streamable HTTP MCP.
MCP endpoint: https://mcp.receiptcal.com/mcp
Use your client's custom MCP connection flow, then authenticate in ReceiptCal's browser page. Never paste passwords or tokens into chat. Client and workspace policies can restrict custom connections. For ChatGPT, use a custom MCP connection only where your plan/workspace supports it; this repository does not bypass directory review.
Add this GitHub marketplace using a current Codex client:
codex plugin marketplace add https://github.com/mihirkavi/receiptcal-plugin.git
codex plugin add receiptcal@receiptcal-pluginsOpen the client's plugin interface, select receiptcal from receiptcal-plugins, and connect your account. The portable plugin.json/mcp.json are canonical; .codex-plugin/plugin.json supports compatible older clients. Host versions differ; consult OpenAI's installation documentation.
claude plugin marketplace add mihirkavi/receiptcal-plugin
claude plugin install receiptcal@receiptcal-pluginsAuthenticate the ReceiptCal server through the host's MCP connection flow. See Claude marketplace documentation.
Related MCP server: MyFitnessPal MCP Server
Node.js 22 or later is required. No backend source checkout is needed.
git clone https://github.com/mihirkavi/receiptcal-plugin.git
cd receiptcal-plugin
npm ci
npm run validate
npm run configure:localThe generated, ignored .mcp.local.json points to the local stdio connector. It is optional; the distributed plugin uses the hosted MCP server. Local account commands include npm run auth -- login and npm run auth -- logout. Enter credentials only in your own terminal. Local auth uses macOS Keychain where available and a permission-restricted local file fallback; environment credentials are supported for controlled development. Do not commit credentials or generated local configuration.
src/capabilities.ts is the typed capability registry; src/server.ts implements MCP tools; src/api-client.ts calls the versioned ReceiptCal APIs; tests use synthetic data and mocked responses. Extend the registry, add schema and behavior tests, run validation, then open a pull request. See CONTRIBUTING.md.
Account writes are prepared first; show the preview and obtain explicit approval before confirm_action. The account, expiry and single-use checks remain enforced. A few recommendation/quality reads may update derived records.
Shopping tools prepare retailer handoffs, never purchases. Nutrition is estimated, with provenance and uncertainty, and is not medical advice.
Camera, HealthKit/Health Connect permission flows and live voice capture remain native app features. The optional local connector can upload receipt files from configured allowed roots; default roots are its working directory and Downloads. Restrict RECEIPTCAL_ALLOWED_FILE_ROOTS to a dedicated test directory.
API protocol versions are compatibility paths, not app version numbers. Some source capabilities depend on service rollout. V6 community recipes remain gated; a tool existing in source does not prove production availability. Do not retry a disabled capability through older routes.
Source tests and protocol checks do not certify a live authenticated journey in every host. See VALIDATION.md.
Service privacy · Service terms · Security reporting
File code bugs with synthetic reproduction steps in GitHub issues. Never include receipts, health details, chats, account identifiers, tokens or screenshots containing personal data. Account-specific reports belong in ReceiptCal's private support flow.
The MIT license covers this repository. It does not grant rights to the private app/backend, customer data, third-party content, or ReceiptCal trademarks. The hosted service retains its own terms and availability limits. This release is not published to npm or the MCP Registry.
This server cannot be deployed
Connect authorized ad accounts to AI assistants for audits, reporting, and approval-gated changes.
Connect your health, fitness, nutrition, sleep, and wearable data to your AI assistant.
Scraps Kitchen gives any AI agent a persistent, household-aware kitchen memory. Unlike generic chatbot recall, Scraps maintains structured cooking data: what's in your fridge (with freshness tracking), who you cook for (with allergens, dietary restrictions, and preferences), your recipe collection (with cook notes and per-diner ratings), your shopping list, and your kitchen equipment. 27 tools across 6 domains let agents read kitchen context, suggest meals that respect dietary safety, update the pantry after cooking, and build a history of what works for your household. Every interaction makes the data richer. Cooking history, preference signals, kitchen awareness = better suggestions next time. All tools work via oAuth and a free scraps.kitchen account.
Household-aware cooking brain: pantry, meal suggestions, dietary safety, recipes, shopping lists.
