ClawTrust
8.1
/ 10
1 evaluations
2.6k Downloads
Overview
Integrate with Swiggy’s MCP servers to search, cart, and place food/grocery (Instamart) orders and make Dineout restaurant bookings in India via a CLI wrapper, with a confirmation-first workflow.
Key Advantages
1.End-to-end Swiggy integration for food delivery, Instamart groceries, and Dineout table bookings from within an MCP/CLI environment.
2.Clear, strongly emphasized safety protocol that forbids auto-ordering and mandates explicit user confirmation with full order preview.
3.Supports cart-based workflows (add/show/clear) for both food and groceries, aligning well with multi-step, LLM-assisted planning.
4.Order logging to a JSON file (memory/swiggy-orders.json) enables basic auditability and history-aware behavior.
5.Works with address aliases like “home”/“office” with guidance to map them from USER.md or ask the user, improving practicality in real use.
Use Cases
- Let an AI agent coordinate and place a team lunch order from nearby restaurants, after interactive menu exploration and cart review.
- Have the assistant assemble weekly Instamart groceries from a recipe or shopping list, then confirm and place the order to a specified address.
- Enable an AI concierge to find restaurants in a given locality and book a Dineout table for a specified date, time, and party size.
- Support budget-conscious ordering by iteratively building a cart and tracking the running total before final confirmation.
- Handle late-night food or urgent grocery needs by searching only currently open options and surfacing estimated delivery times before ordering.
Evaluation Scores
8.1
/ 10
Reliability
7.0
Functionality
8.5
Usability
8.2
Safety
8.8
Performance
7.5
Compatibility
8.0
Based on 1 evaluation · Latest: 3/20/2026
Download Trend
Loading...
Evaluation History (1)
8.1/103/20/2026▼
OS: linux-arm64LLM: anthropic/claude-opus-4.6
**Quick judgment**: This is a strong, domain-focused skill for Swiggy users in India, covering food delivery, Instamart groceries, and Dineout restaurant bookings with a well-documented, confirmation-first flow. It’s well-suited for AI agents that act as a food/grocery concierge, provided the environment has Node.js, `mcporter`, and CLI support.
**Key strengths**
- Rich Swiggy coverage: search, menus, carts, ordering, Instamart, and Dineout bookings.
- Explicit safety model: never auto-order; always preview items, quantities, prices, total, address, and ETA; requires a clear YES before using `--confirm`.
- Example-driven workflows for food, grocery, and restaurant booking scenarios, which map cleanly to multi-step agent plans.
- Order logging to JSON offers basic traceability for what the agent has ordered.
**Main risks / limitations**
- **Irreversible COD orders**: Only Cash on Delivery is supported and orders cannot be cancelled; any confirmation bug or misunderstanding can directly cost the user money or cause inconvenience.
- **Geographic scope**: Swiggy is India-only; functionality is largely useless outside supported regions.
- **Dependency chain**: Requires `mcporter`, the Swiggy MCP, OAuth auth flow, and Node.js; any break in that chain can degrade reliability.
- **Session conflicts**: The skill warns not to open the Swiggy app simultaneously due to possible session issues.
**Recommended scenarios**
- AI assistants for users in India that frequently use Swiggy for food or groceries and are comfortable with COD.
- Team or household ordering workflows where the AI aggregates preferences, builds a cart, and then seeks one final human confirmation.
- “Food/grocery concierge” or “dining concierge” agents that combine menu exploration, budget checks, and logistics (address/ETA) before placing any order.
Given the irreversible nature of COD orders, this skill should only be enabled for agents that strictly respect the confirmation contract and clearly communicate totals, address, and ETA before executing any `--confirm` command.
Comments (0)
No comments yet. Be the first!