Skip to main content
The NexusFeed MCP server exposes five tools across two sub-servers (nexusfeed-ltl and nexusfeed-abc) and two prompt templates. Each tool wraps one of the REST endpoints documented in the API Reference — the MCP payloads and JSON responses are identical.

LTL sub-server (nexusfeed-ltl)

What it does: Returns all ten supported LTL carriers with their SCAC codes, full names, headquarters, and coverage.When to call it: At the start of a conversation when the agent needs to know which carrier codes are valid. Also useful as a “what’s available” query.Underlying endpoint: GET /v1/ltl/carriersExample agent prompt: “What LTL carriers can NexusFeed look up?”
What it does: Returns the current and historical fuel surcharge rate for one carrier, with optional history depth (1-26 weeks). Includes the DOE diesel price (on carriers that publish it) and a _verifiability block.When to call it: Whenever the user asks for a fuel surcharge rate, wants to verify an invoice, or wants to compare carriers. This is the most-called tool in the LTL sub-server.Parameters:
  • carrier (required): SCAC code — ODFL, SAIA, ESTES, ABF, RLC, TFORCE, XPO, SEFL, AVERITT, FEDEX
  • weeks (optional, default 4): Integer 1-26
Underlying endpoint: GET /v1/ltl/fuel-surchargeExample agent prompt: “What’s ODFL’s fuel surcharge this week?” or “Compare ODFL, Saia, and Estes fuel surcharges for the last 4 weeks.”
Status: Coming soon. Returns 404 COMING_SOON today.What it will do: Return the accessorial fee schedule for a carrier (liftgate, residential delivery, re-delivery, inside delivery, limited access, notification, appointment).Underlying endpoint: GET /v1/ltl/accessorials

ABC sub-server (nexusfeed-abc)

What it does: Searches for liquor licenses in a given state by trade name, owner name, or address. Returns a list of matching records with status, expiration, license type, and a _verifiability block.When to call it: Whenever the user has a business name or owner but not a specific license number. At least one of dba_name, owner_name, or address must be provided.Parameters:
  • state (required): Two-letter code — CA, TX, NY, FL, IL
  • dba_name (optional): Trade name, partial match supported
  • owner_name (optional): Holder name, partial match supported
  • address (optional): Street/city filter
Underlying endpoint: GET /v1/abc/searchExample agent prompt: “Search California ABC for all Trader Joe’s licenses.” or “Does this New York bar named ‘Employees Only’ have a current liquor license?”
What it does: Direct lookup of a single license by its state-issued number. Returns one record or 404 if not found.When to call it: When the user has a specific license number and wants precise verification — an insurance binding check, a compliance audit trail, a due-diligence lookup.Parameters:
  • license_number (required): State-issued license number
  • state (required): Two-letter code matching the license
Underlying endpoint: GET /v1/abc/license/{license_number}Example agent prompt: “Look up California ABC license 47-123456.”

Prompt templates

Both sub-servers ship with prompt templates that agents can use for common workflows.

ltl_fuel_surcharge_audit

Instructs the agent to fetch a carrier’s current rate and check the response against a user-supplied invoice rate, flagging discrepancies.

abc_license_compliance_check

Instructs the agent to look up a specific license, verify it is ACTIVE and not expired, and return a pass/fail with the source evidence URL.

How tools behave under error

  • UNAUTHORIZED (401) — surface a clear “API key invalid” message; don’t retry.
  • PRODUCT_NOT_ENABLED (403) — tell the user they need to subscribe to the other RapidAPI listing.
  • NOT_FOUND (404) — report the carrier/state/license as unsupported.
  • SOURCE_UNAVAILABLE (503) — the upstream carrier or agency site failed. Retry after a delay, or report the outage.
  • CAPTCHA_SOLVER_NOT_CONFIGURED (503) — only happens on self-hosted deployments missing a 2Captcha key; production never emits this.

Gate on verifiability

Agents should always check _verifiability.extraction_confidence before using a returned value in a billing decision or compliance statement. The recommended threshold is >= 0.90. See the Verifiability contract for the full story.