measure-mcp

fba_measure_package

Measure a packaged box or bag from 1–5 3D scans to get dimensions in inches, girth, volume, and weight, returning data ready for Amazon FBA size tier and fee analysis.

Instructions

Measure the PACKAGED unit (the box or bag as it ships) from 1 to 5 scans of the same package.

Inputs: path = one scan file, or paths = a list of 2 to 5 scans of the same pack (combined by median per sorted edge). A scan is a 3D scan export (a Scaniverse, Polycam or 3D Scanner App OBJ or PLY mesh or point cloud; a .zip holding the .obj plus its .mtl and texture is fine) of the package on a table, a metric depth file (HEIC, TIFF, 16-bit PNG, NPZ) or an iOS capture bundle (folder or .zip); rgb_path is an optional companion photo for a single depth file. For OBJ/PLY the file's units (m, cm, mm or inches) and up axis are detected from the table and plausible sizes, and notes say what was assumed; if that is wrong or ambiguous, pass scan_units (m|cm|mm|in: what the FILE is written in; Scaniverse writes METERS, and units="cm" works as an alias) and up_axis (x|y|z, signs ok). The package is the object nearest the middle of the scan. weight is the user's own scale reading of the packed unit (a scan cannot weigh): a number (POUNDS unless weight_unit says otherwise) or "22 oz" / "640 g"; weight_sigma_oz is the scale's 1-sigma error in ounces. bbox [x0, y0, x1, y1] or point [x, y] pick the package out of the scene; they are in PHOTO pixels when the scan carries a photo (HEIC or a capture bundle with its colour image), otherwise in depth-image pixels (the result's pixel_space says which). reference is only used when the depth is relative rather than metric: {"type": "credit_card", "corners": [[x,y] x4]} or {"type", "bbox"} in the same pixel space (see measure_list_reference_objects).

Returns dims_in {longest, median, shortest} in INCHES with 1-sigma sigma_in per edge (LiDAR is never reported better than 0.2 in), dims_cm, length_plus_girth_in, volume_cubic_in, unit_weight (user-supplied), confidence (high/medium/low), quality_flags (thin_item_unreliable, low_coverage, soft_or_irregular_surface, scan_disagreement, monocular_only, relative_depth_not_metric, assumed_intrinsics), capture_advice and fba_args, ready to pass to fba_analyze. Relative, monocular or assumed-intrinsics depth is never high confidence. Items thinner than 1.5 in cannot be certified by a depth scan.

This is an estimate for packaging decisions; Amazon measures the unit itself and its measurement governs. detail="full" adds per-scan results and the assumptions list.

Input Schema

TableJSON Schema

Name	Required	Default
`bbox`	No
`path`	No
`paths`	No
`point`	No
`units`	No	both
`detail`	No	summary
`weight`	No
`up_axis`	No	auto
`rgb_path`	No
`reference`	No
`scan_units`	No	auto
`weight_unit`	No
`weight_sigma_oz`	No

Output Schema

TableJSON Schema

Name	Required	Description	Default
No arguments

Schema Changelog

Changes observed during successful MCP inspections.

First observedv0.1.0Oct 16, 2026

TDQS

A4.7/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden and does so richly: it states reliability ceilings ('LiDAR is never reported better than 0.2 in'), a hard limitation ('items thinner than 1.5 in cannot be certified'), confidence rules for relative/monocular/assumed-intrinsics depth, that a scan cannot weigh, and that Amazon's own measurement governs. These are exactly the traits an agent needs before calling.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

It is long and dense, but front-loaded with the purpose first and then explicitly partitioned into 'Inputs' and 'Returns' plus a closing caveat, so an agent can navigate it. Given 13 parameters at 0% schema coverage and zero annotations, the length is largely earned rather than padding.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 13-parameter tool with an output schema and no annotations, the definition covers input formats, unit detection, filtering, the return payload, and confidence/limitation semantics. Nothing an agent needs in order to call it correctly is missing, and the return summary adds interpretive value beyond the schema.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 0% across 13 parameters, yet the description supplies deep semantics for nearly all of them: file/unit/up-axis detection and override aliases ('Scaniverse writes METERS, units="cm" works as an alias'), weight formats and default units, pixel-space rules for bbox/point, the reference schema, and detail="full" behavior. Only the 'units' output-formatting parameter goes unmentioned.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The opening sentence gives a specific verb and a sharply scoped resource: 'Measure the PACKAGED unit (the box or bag as it ships) from 1 to 5 scans of the same package.' The parenthetical and the multi-scan framing distinguish it from generic siblings such as measure_object and kitchen_measure_vessel without the agent needing to open a schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description gives strong in-scope guidance (1 scan via path vs 2-5 same-pack scans via paths, when to override scan_units/up_axis, when reference is needed) and cross-references measure_list_reference_objects. It stops short of naming when to prefer this over measure_object/kitchen_measure_vessel, so the when-not case is left to inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.