Public assessment

SoapyRED/freightutils-mcp (freightutils-mcp)

freightutils-mcp · v2.12.1 · scanned Aug 2, 2026

What changed in the harness

Selection accuracy 100→100, token cost down 0%, unconfirmed writes 100%→100%.

Category breakdown

Where the score comes from.

Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.

01Safety

30.0 / 30

30.0 out of 30

02Legibility

27.3 / 30

27.3 out of 30

03Economics

4.5 / 20

4.5 out of 20

04Discoverability

15.6 / 20

15.6 out of 20

Highest-impact fix

Estimated gain +15 points

Reduce the tool-definition and task token payload

Consolidate overlapping tools and trim repetitive schema text while preserving behavior, parameter, and return semantics.

Description evidence

Defects and rewrites.

2 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.

Tool	Defect types	Suggested rewrite
`container_lookup`	params_unexplained	Get ISO shipping-container specifications, with optional load-fit maths. Covers 10 types: 20ft/40ft standard, 40ft and 45ft high-cube, 20ft/40ft reefer, 20ft/40ft open-top and 20ft/40ft flat-rack. Provide type as a slug (e.g. "20ft-standard", "40ft-high-cube") for one container's record; omit it to list all 10. Add item dimensions — item_length_cm, item_width_cm and item_height_cm (all three, in centimetres, giving one item's length/width/height) — plus optional item_weight_kg and item_quantity, to also compute how many such items fit. Behavior: read-only reference data with per-record provenance (sources, audited_at, decision_rationale); an unknown type errors with the valid slug list. Fit calculations are geometric best-effort — they do not model load distribution, securing or mixed cargo. Rate-limited (anonymous use: 25 requests/day per IP): a 429 error body carries retry_after_seconds and a Retry-After header — back off and retry, or call get_subscribe_link for higher limits. Returns: the container record — internal/external/door dimensions (cm), capacity_cbm, tare_weight_kg, max_gross_kg, max_payload_kg and euro/GMA pallet counts — under result, plus confidence, _source and citation (the FreightUtils v1 response envelope). Limitations: manufacturer-typical specs, provenance pending independent verification (the envelope's provenance_status says so) — actual equipment varies by lessor and line; confirm against the carrier's equipment guide. Related: validate (checks a container NUMBER's ISO 6346 check digit — not specs), cbm_calculator / consignment_calculator (the cargo volume to fill it), uld_lookup (the air-freight equivalent).
`pallet_fitting_calculator`	params_unexplained	Calculate how many identical boxes fit on a pallet: boxes per layer (trying 90-degree rotation when allowed), layer count within the max height, totals, volume utilisation and weight capping. Provide pallet_length_cm, pallet_width_cm and pallet_max_height_cm for the pallet's footprint and stack limit, and box_length_cm, box_width_cm and box_height_cm for the single box's length, width and height (all in centimetres) being packed onto it. Behavior: deterministic geometric packing of one box size in aligned rows and columns — it does not model interlocked or mixed-orientation patterns; weight_limited reports when max_payload_kg caps the count below the geometric fit; pallet_deck_height_cm defaults to 15. Missing or non-positive dimensions error naming the parameter. Rate-limited (anonymous use: 25 requests/day per IP): a 429 error body carries retry_after_seconds and a Retry-After header — back off and retry, or call get_subscribe_link for higher limits. Returns: boxes_per_layer, layers, total_boxes, orientation, boxes_per_row/col, usable_height_cm, utilisation_percent, total_box_volume_cbm, wasted_space_cbm and the weight fields under result, plus confidence, _source and citation (the FreightUtils v1 response envelope). Limitations: a theoretical best-effort fit — real stacking obeys carton strength, overhang and load-stability rules it does not model. Related: ldm_calculator (pallets into trailer length), vehicle_lookup (pallet capacity per vehicle), container_lookup (pallets into containers).

Selection evidence

Confusable tool pairs.

8 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.

Tool A	Tool B	Confidence	Why they collide
`cbm_calculator`	`consignment_calculator`	high	A vague task like 'calculate the cubic metres for my shipment' is genuinely ambiguous when the shipment has multiple different item types — consignment_calculator computes per-line CBM as part of its output too, so an agent could pick either the single-purpose cbm_calculator or the multi-line consignment_calculator depending on how it parses 'shipment' vs 'piece'.
`ldm_calculator`	`consignment_calculator`	high	consignment_calculator's road mode returns loading metres per line and in totals, duplicating ldm_calculator's core output; a task like 'work out the loading metres for these mixed pallets' could plausibly be routed to either tool since both legitimately answer it.
`airport_lookup`	`nearest_airport`	medium	A task like 'find the airport near London' is ambiguous — it could mean a name/city text search (airport_lookup with query='London') or a coordinate-based proximity search (nearest_airport), and the agent must infer whether the user has coordinates or just a place name in mind.
`airline_lookup`	`airport_lookup`	medium	Both accept a bare 'code' lookup and share the IATA-code concept; a task like 'look up code LHR/EK' without specifying carrier vs airport could lead an agent to pick the wrong tool, especially since both use similarly-shaped 2-3 char code parameters.
`cbm_calculator`	`ldm_calculator`	medium	Both are 'how much space does my cargo take' calculators (volume vs trailer loading metres) fed by piece dimensions; a generic task like 'how much room will these pallets take on the truck' could be misrouted to cbm_calculator instead of the road-specific ldm_calculator.
`uld_lookup`	`vehicle_lookup`	medium	Both return equipment specifications (dimensions, payload/weight capacity, deck info) for cargo-carrying units, differing only by air (ULD) vs road (vehicle/trailer); a mode-agnostic task like 'look up the payload capacity of this loading unit' could be sent to either.
`chargeable_weight_calculator`	`ldm_calculator`	low	Both are freight-billing-adjacent calculators from piece dimensions ('calculate freight cost basis for my shipment'), but they serve different modes (air chargeable weight vs road loading metres) so confusion only arises if the task omits the transport mode.
`chargeable_weight_calculator`	`pallet_fitting_calculator`	low	Both take box/piece dimensions plus weight and produce a weight-capped figure ('calculate the effective weight/capacity for these boxes'), which could momentarily mislead an agent despite serving unrelated purposes (air billing weight vs boxes-per-pallet count).

Compare the field

One score is useful.
The evidence makes it actionable.

Back to the leaderboard