avasfood-mcp
A standard Python MCP server for AVAS Food: browse menus, build and quote carts, complete bank-transfer ordering with explicit confirmation, track deliveries, and analyze delivery spending.
README
AVAS Food MCP — client-agnostic food ordering
A standard Python MCP server for AVAS Food: browse menus, build and quote carts, complete bank-transfer ordering with explicit confirmation, track deliveries, and analyze delivery spending.
The server uses standard MCP over stdio and has no Hermes runtime dependency. It works with Claude Code, OpenAI Codex CLI, Cursor, Gemini CLI, OpenCode, VS Code/Copilot, Windsurf, Hermes Agent, and other MCP-compatible clients.
Quick start
git clone https://github.com/Rayyan011/avasfood-mcp.git
cd avasfood-mcp
./scripts/setup.sh
Then choose a client configuration from docs/CLIENTS.md. Configuration templates live in configs/.
Remotion demo
The repository includes a 36-second, 1080p Remotion product demo under demo/.
cd demo
npm install
npm run render
The rendered MP4 is written to demo/out/avasfood-mcp-demo.mp4.
Tools
Existing tools:
get_client_metadatalist_restaurantsget_restaurantsearch_menu_itemsaccount_status,start_phone_login,verify_phone_loginlist_delivery_addresses
Ordering tools:
get_menu— authenticated merchant menu with zone, coordinates, and address contextadd_to_cart,update_cart_item,remove_cart_item,view_cart,clear_cart— local-only, single-merchant canonicalCartRowstoragequote_cart— callsCalculationSalesOrderand parses totals, payment modes, and bank detailsstage_receipt— validates/copies a PNG or JPEG locally; does not uploadupload_receipt— uploads 500,000-character base64 chunks only after the exact confirmation phraseprepare_order— freezes the current cart/quote into an immutable, redacted preview and one-time tokensubmit_order— requires that exact token, validates, then callsSaveSalesOrderoncetrack_deliveries— live active-order, rider, and delivery statedelivery_spending_report— private spending totals by fees, restaurant, and inferred food type
Portable agent guidance
AGENTS.md— vendor-neutral behavior and safety rulesCLAUDE.md— Claude Code entry pointGEMINI.md— Gemini CLI entry point- MCP prompts:
food_assistantanddelivery_budget_analyst - Optional skills under
skills/for clients with compatible skill loaders
Hermes users may additionally install the optional skill packages:
mkdir -p ~/.hermes/skills/avasfood
cp -R skills/avas-food-assistant ~/.hermes/skills/avasfood/
cp -R skills/food-delivery-budgeting ~/.hermes/skills/avasfood/
The tools do not require those skills; they improve conversational orchestration only.
Safety model
- Local state is stored under
~/.avasfood-mcpwith private directory/file permissions. - Carts accept only one merchant at a time.
- Bank transfer is payment mode
4; a receipt GUID is required before preparation. - Receipt upload requires exactly:
I CONFIRM UPLOAD OF THIS BANK TRANSFER RECEIPT. - Preparing does not submit. Submission requires the exact generated token and an unchanged cart.
SaveSalesOrderis never automatically retried. Before the call, the order is marked in-flight; transport failures are persisted as ambiguous and users must inspect order history.- Tool results redact addresses/names and expose only the last four phone digits. Secrets are not stored in this repository.
- Spending history is written privately to
~/.avasfood-mcp/spending-history.jsononly after a successful MCP-submitted order. It does not claim to include older AVAS orders.
Typical flow
- Link the account (
start_phone_login,verify_phone_login) if needed. - Select an address/zone and call
get_menu. - Build and inspect the local cart.
- Call
quote_cart. - If using bank transfer (
4), callstage_receipt, inspect the staged metadata, then separately callupload_receiptwith the exact phrase. - Call
prepare_orderand carefully inspect its preview. - Separately call
submit_orderwith the returned token.
Run
cd /Users/rayyu/personal/avasfood-mcp
python3 -m venv .venv
.venv/bin/pip install -e .
.venv/bin/python -m avasfood_mcp.server
Optional environment variables:
AVAS_ACCESS_TOKEN— approved backend token, if requiredAVAS_API_BASE— owner-approved API overrideAVAS_ENVIRONMENT— defaults toLIVEAVAS_STATE_DIR— local private state override
Tests
Tests mock _post; they do not call AVAS production calculation or write endpoints.
.venv/bin/python -m unittest discover -s tests -p 'test_*.py' -v
Example stdio configuration
{
"mcpServers": {
"avasfood": {
"command": "/absolute/path/avasfood-mcp/.venv/bin/python",
"args": ["-m", "avasfood_mcp.server"]
}
}
}
See docs/CLIENTS.md for client-specific locations and commands.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.