Loading ToolBench
MCP server for the Warp freight API, covering the full booking lifecycle: mode compare, quote (van/box-truck/FTL/LTL), batch quote, book, batch book, multi-stop FTL, track, events, invoices, documents, lane history, quote history, payment status, locations, load templates, in-chat login, status, analytics. Renders inline quote, batch-quote, batch-book, and shipments cards in Claude and ChatGPT.
This is a well-structured freight logistics MCP server with 26 tools covering quoting, booking, tracking, and account management. Strengths: all 26 tools have explicit names following verb_noun pattern (compare_modes, van_quote, book, track, etc.); all have descriptions (ranging 50-200+ chars); input schemas are present with typed parameters and descriptions for most tools; output handling includes structured card rendering for UI-capable clients. Weaknesses: (1) Some parameter descriptions lack explicit format/constraint details (e.g., 'origin_zip' says '5-digit US ZIP code' but could specify regex pattern or validation guidance); (2) No documented output schemas for many tools, responses are rendered as cards but the JSON structure is not formally specified; (3) Error handling appears minimal in visible code, no evidence of recovery guides, error classification, or actionable error messages; (4) Some tools combine multiple concerns (e.g., multistop_book both quotes and books); (5) Tool descriptions are good but could be more explicit about when to use each variant (van_quote vs box_truck_quote vs ftl_quote vs ltl_quote distinctions could be clearer); (6) No evidence of pagination parameters on list_* tools (list_bookings, list_documents, list_invoices, list_events); (7) Response field naming must chain properly but is not fully visible in provided code. Overall, this is a mature, production-oriented server with strong naming and description discipline, but lacks formal output schema documentation and robust error handling patterns.
Get aggregated usage analytics for the authenticated user. Returns quote count, booking count, shipment volume, and spending trends.
Confirm N quoted lanes in one call with a single progress card instead of N noisy per-row calls. Returns booking status for each row (Booked/Failed), tracking numbers, and a consolidated progress view.
Price an entire spreadsheet of lanes in one call, rendered as a single card. Returns all results (succeeded and failed) with quote_ids for bookable lanes.
Confirm a quoted shipment and create a booking. Requires a quote_id from a prior quote call. Returns tracking number, shipment details, and booking confirmation.
Get a quote for box truck freight service. Prices a load up to 12 pallets for mid-range delivery via 26' box truck.
Output schemas not formally documented. Tools return structured HTML card widgets and JSON, but the JSON response structure (fields, types, cardinality) is not documented in tool definitions. LLMs cannot plan downstream tool calls without knowing what fields to expect.
List tools (list_bookings, list_invoices, list_documents, list_events, list_locations, list_load_templates, quote_history) lack pagination parameters (limit, offset/cursor). Without pagination, large result sets can exhaust context or cause timeouts.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 67 | 2025-06-18+ | v2 |
Cancel a booked shipment. Only works for shipments not yet picked up. Returns cancellation status and any applicable refund information.
Compare all available freight modes (LTL, FTL, van, box-truck) for a single lane in parallel. Returns a decision-complete recommendation with cost-vs-transit trade-off, pricing every eligible mode and explicitly marking unavailable ones with the reason why.
Get a quote for FTL (Full Truckload) freight service. Prices a full 53-foot trailer with up to 26 pallets.
Retrieve a specific shipment document (BOL, proof-of-delivery, etc). Returns document content and metadata.
Get detailed invoice information including charges, taxes, and payment status. Returns a PDF-ready invoice document.
Get historical quote and booking data for a specific lane. Shows previous rates, modes used, and transit times to help with pricing trends and planning.
List all shipments for the authenticated user. Returns a TMS-style detail view where each shipment can be expanded to see pickup/delivery detail, freight info, and tracking links. Renders as an interactive card with progressive carrier spread fill-in.
List all shipment documents (BOLs, proof-of-delivery, insurance docs). Returns document type, shipment reference, and download links.
Get timeline events for a shipment. Returns pickup, delivery, exception, and status change events with timestamps.
List all invoices for the authenticated user. Returns invoice numbers, dates, amounts, and line-item details.
List saved load templates for the authenticated user. Templates capture recurring freight profiles (dimensions, weight, commodity) for quick re-quoting.
List pickup and delivery locations (origins/destinations) for the authenticated user. Useful for identifying commonly-used facilities and validating coverage.
In-chat login flow. Displays a browser-based login dialog for users without a prior API key. Returns an API key on successful authentication.
Get a quote for LTL (Less Than Truckload) freight service. Splits the fast Warp rate from slower carrier comparison; user sees a price in 1-2 seconds. Supports optional dimensions, commodity type, freight class, and hazmat flags.
Get authenticated user profile information. Returns account name, email, company, and subscription tier.
Quote and book one truck across 3+ stops (multi-stop FTL). Prices and confirms a single vehicle for multiple pickup and delivery locations.
Get payment status and balance information for the authenticated user. Shows account credit, outstanding charges, and payment history.
Get all historical quotes for the authenticated user. Returns quote IDs, pricing, modes, and expiration status.
Get system status and health check. Returns server status, API availability, and any ongoing incidents or maintenance.
Get real-time tracking information for a shipment by booking_id. Returns current location, estimated delivery, and tracking history events.
Get a quote for van freight service. Prices a load up to 3 pallets for expedited delivery via cargo van.
Parameter format constraints are described in English ('5-digit US ZIP code', 'ISO 8601 date') but not as formal JSON Schema patterns or enums. LLMs cannot validate input format without regex patterns or enum declarations.
multistop_book combines quoting and booking in one tool. Per pattern:tool, each tool should do exactly one thing. This forces the agent to commit to booking without the option to review quotes separately.
No evidence of error handling patterns in visible code. Tools should return recovery guidance (e.g. 'Coverage gap for this lane. Try searching nearby ZIP codes.'), error classifications (retryable vs user-fixable vs fatal), and actionable messages. Current approach appears to rely on raw API responses.
Quote variants (van_quote, box_truck_quote, ftl_quote, ltl_quote) are separate tools. Descriptions mention mode differences but could more explicitly guide when to use each (e.g. 'Use van_quote for 1-3 pallets with fast delivery; use ltl_quote for 1-26 pallets with slower, cost-optimized transit'). compare_modes already does this, but the individual tools don't clarify their niche.
Irreversible operations (book, batch_book, multistop_book, cancel_shipment) lack confirmation or dry-run support. Agents can accidentally commit to bookings without review. Consider adding a confirm_before_execute pattern or a dry_run parameter to preview the operation.