Engineering code-table data with citations: standard, edition, table, row and column.
Scored across 10 tools
Most tools have clearly distinct purposes: describe_dataset (schema), find_dataset (search), list_keys (row identifiers), lookup_value (single value) and get_pointer (location only) form a readable progression, and the domain-specific tools (get_aisc_shape, get_pipe_dimensions, get_skey_map, get_fixture_example, verify_connector) are unmistakable. The main soft spot is get_pointer vs lookup_value in pointer mode (which returns a citation with null value), plus mild overlap between find_dataset and list_keys for enumerating what exists.
Every tool is snake_case with a leading verb (describe_, find_, get_, list_, lookup_, verify_) followed by a noun, and the pattern is applied uniformly across both generic dataset tools and domain-specific ones. No camelCase or mixed-style names appear.
Ten tools is well within the ideal band and each one earns its place: four generic dataset tools, four domain-specific lookup tools, and two identity/support tools. Nothing looks redundant or padding-like.
The surface covers the full lookup lifecycle (discover, describe, enumerate keys, fetch value, fall back to a pointer) plus concrete cover for pipe dimensions, AISC shapes, SKEY maps and fixtures. Minor gaps: no way to enumerate all datasets without a keyword, no batch/multi-row lookup, and shapes are only addressable by designation, not by property search.
Describe one dataset: the key fields a lookup needs, the value columns with units and which need a temperature, the citation (standard, edition, table), whether lookups return a value or a pointer, the tier at which values are returned, and the daily lookup limits. Returns structure only, no table values.
Inputs: dataset_id (required). Returns: key, caps, mode, title, +6 more
Search the engineering code-table datasets by keyword (standard, table, quantity, material). Returns up to 10 matches, each with its id, title, standard, edition and table, whether lookups return a value or a pointer (table, row and column only), and the tier at which values are returned.
Inputs: query (required), standard. Returns: results
Return the dimensions and section properties of one AISC shape by designation (e.g. W8X31, HSS6X6X3/8, PIPE4STD, L4X4X1/2): weight, area, depth, flange and web dimensions, I, S, Z, r, J and Cw as applicable. Source: AISC Shapes Database v16.0; L, HSS and Pipe computed from nominal geometry.
Inputs: designation (required). Returns: family, source, properties, attribution, +1 more
Return one worked example case for a Piping Toolset plugin skill: the inputs, the expected outputs and the pass/fail status from the hand-checked calculation fixtures, with case names removed. case selects among the skill's cases (0-based; the result gives the count).
Inputs: skill (required), case. Returns: tol, input, skill, expect, +4 more
Return the outside diameter and nominal wall thickness (in) for a nominal pipe size per ASME B36.10M / B36.19M: one schedule with its wall and inside diameter when schedule is given (e.g. 40, STD, XS, 80, 10S), otherwise the walls of every schedule listed at that size.
Inputs: nps (required), schedule. Returns: od, nps, mode, unit, +6 more
For a quantity whose value this server does not return (for example PVP2011-57100 nozzle allowables, ASCE 7-22 SDS, Ip, CAR and Rpo, AISC Fy/Fu, ACI anchor product values, API 617/611/612 and NEMA SM 23 coefficients, API 650 Annex P coefficients, and any dataset currently returned as a location only), return where to read it: standard, edition, table or figure, the row key and columns, and the inputs needed. Never returns a value.
Inputs: quantity (required), standard (required). Returns: pointers
Return the Isogen-style SKEY vocabulary Piping Toolset uses (component prefixes, end-condition suffixes, fitting and support keys) with scope 'isogen', or the PCF record set, substitutions and default SKEYs that import cleanly into CAESAR II with scope 'caesar-safe'. component narrows the result to one group, e.g. 'valve_prefix' or 'safe_substitutions'.
Inputs: scope (required), component. Returns: about, scope, groups
List the key identifiers of a dataset's rows (for example spec, grade, class, size, and the row position line) so lookup_value can address exactly one row. Returns identifiers only, never table values, at most 50 per call; prefix filters on the first key field. For pointer datasets it returns the key schema without rows.
Inputs: dataset_id (required), limit, prefix. Returns: keys, mode, truncated, dataset_id, +1 more
Return one value from one row and one column of a code-table dataset, with its citation (standard, edition, table, row, column), unit and the tabulated entries used. Temperature columns take T in °F; allowable-stress and flange-rating columns interpolate linearly between the two bracketing tabulated temperatures and never extrapolate. Datasets in pointer mode return the citation with value null. The key must identify exactly one row; use list_keys for identifiers.
Inputs: key (required), dataset_id (required), column. Returns: mode, text, unit, value, +7 more
Report whether the caller is signed in, the caller's tier, how many datasets return values at that tier, and the remaining daily lookups when the server tracks them. Returns no email, name or account id.
Inputs: no named parameters. Returns: tier, signed_in, caps_remaining, datasets_available, +1 more
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Source-backed public tables with citations, claim checks, and reviewable corrections.
Read-only sourced engineering materials data: every value with unit, conditions and source.
Cited product-compliance ground truth for AI agents. Never generates; always cites.
Shared, versioned tables with units, links and SQL. Agents read what your data means.
Add one secure layer between your agents and this server.
