analyze_beam
ChatGPTWhen beam details are missing, proceed with reasonable, clearly stated assumptions and calculate a preliminary result instead of routinely blocking on clarification. Preserve every supplied input. Briefly state the assumptions before calling a calculation tool, and record each assumed value or arrangement in inputAssumptions so it remains visible with the result. With no unit context, use metric units; for an otherwise unspecified simple beam, use pinned/roller end supports, a downward point force at midspan and excluded self-weight, explicitly labeling whichever of these were assumed. Use established context on follow-ups, retaining unconfirmed assumptions and updating or removing assumptions the user changes or confirms. Ask a concise question only when contradictory inputs prevent a coherent model, or when the requested behavior is outside the supported beam analysis. Do not guess design limits, claim code compliance or certify construction safety. Certification, member capacity, buckling, bracing, connections, load combinations and design-code checks are outside this plugin. When asked for any of these, explain the limitation without offering a preliminary code-style check, a code pass/fail or professional approval as a substitute. You may offer only the supported elastic analysis or catalog properties, clearly stating that these do not establish construction safety. The plugin cannot access saved account models, account settings, credentials or conversation history, and cannot send data to arbitrary destinations. Catalog-only lookups need no beam assumptions. Calculate a general linear elastic beam with multiple supports, section regions, point loads, linearly varying loads and applied moments. Return actual solver reactions, shear, moment, slope, deflection and available stresses. Units and self-weight basis are required. Omit sections for reactions, shear and moment under a stated constant-EI assumption, including indeterminate beams. Numerical E/I are needed for slope and deflection; record any assumed material or section properties in inputAssumptions. Changes of stiffness along an indeterminate beam require the relative EI distribution for force results; retain supplied section regions. The resultId is a bearer reference to an in-memory result snapshot retained for up to 30 minutes. This data tool does not render a card: after the final calculation call show_beam_results with resultId. For multiple candidate steel sections use compare_steel_sections instead of repeated analysis/display calls. Never claim member capacity or code compliance.
compare_steel_sections
ChatGPTWhen beam details are missing, proceed with reasonable, clearly stated assumptions and calculate a preliminary result instead of routinely blocking on clarification. Preserve every supplied input. Briefly state the assumptions before calling a calculation tool, and record each assumed value or arrangement in inputAssumptions so it remains visible with the result. With no unit context, use metric units; for an otherwise unspecified simple beam, use pinned/roller end supports, a downward point force at midspan and excluded self-weight, explicitly labeling whichever of these were assumed. Use established context on follow-ups, retaining unconfirmed assumptions and updating or removing assumptions the user changes or confirms. Ask a concise question only when contradictory inputs prevent a coherent model, or when the requested behavior is outside the supported beam analysis. Do not guess design limits, claim code compliance or certify construction safety. Certification, member capacity, buckling, bracing, connections, load combinations and design-code checks are outside this plugin. When asked for any of these, explain the limitation without offering a preliminary code-style check, a code pass/fail or professional approval as a substitute. You may offer only the supported elastic analysis or catalog properties, clearly stating that these do not establish construction safety. The plugin cannot access saved account models, account settings, credentials or conversation history, and cannot send data to arbitrary destinations. Catalog-only lookups need no beam assumptions. For property-only comparisons use show_steel_sections, which needs no beam model. This tool compares elastic beam responses. Compare 2–6 exact catalog candidates on the same beam in one call. Replaces selected section regions with each candidate while retaining supports, loads, E, units and self-weight basis. Returns verified elastic stress/deflection, depth and weight, with checks ONLY against explicitly supplied user limits. Defaults to an unranked shortlist; do not invent selection criteria. Rankings cover compared candidates only and do not establish member suitability or code capacity. The returned resultId is a bearer reference to an in-memory comparison retained for up to 30 minutes. Call show_beam_results once with that resultId; do not analyze or display each candidate separately.
find_steel_alternatives
ChatGPTFind deterministic nearby alternatives to an exact catalog section: lighter/heavier in its depth series, another similar Ix option, and a neighboring depth series with similar weight when available. Returns exact references and reasons, keeping the same catalog edition and family. No beam inputs are needed. Then call show_steel_sections with the original and chosen alternatives. These are property alternatives, not design recommendations.
get_steel_section
ChatGPTRetrieve a specific steel section using the exact library and slug from search_steel_sections. Returns geometric properties in metric and imperial units. Material stiffness E must be specified separately for analysis.
search_steel_sections
ChatGPTFind catalog steel sections across all five regions by default, or a specified region, designation, family, depth series or geometric-property filters. Sort all matching records by relevance, lowest weight, smallest depth or highest Ix before pagination. Missing ranking values are reported; this is geometric ranking, not whole-catalog beam optimization. For a section lookup or property comparison, call show_steel_sections once with the final exact references; no beam inputs are needed. Returns edition-scoped references and properties in both unit systems. This is optional for general beam analysis and does not check member capacity or suitability.
show_beam_results
ChatGPTDisplay one final analysis or comparison returned by analyze_beam or compare_steel_sections. Pass the exact resultId. Set diagram to the result the user asked to see, for example deflection for a deflection question or shear for a shear question; otherwise omit it. Never supply invented result values. For a comparison, the card shows the verified shortlist and lets the user inspect any candidate. Call once after the final calculation. Results expire after 30 minutes or a service restart; recalculate if expired. For a preliminary calculation, state the assumed inputs before calling the tools and invite corrections after the card. The interactive card already answers the request with drawings, inputs, tables and results. After displaying it, use at most two short sentences for one useful interpretation, tradeoff or missing input. Do not repeat numerical lists, tables, formulas, input summaries, ASCII drawings or another schematic unless the user explicitly asks for them. A short explanation may quote a number when it is needed to explain the observation. For a simple lookup with nothing useful to add, a brief confirmation is enough. In a client without a rendered card, use the data-tool results to provide the requested answer.
show_steel_sections
ChatGPTShow one catalog section or compare 2–6 sections by dimensions, weight and geometric properties, with schematics, unit switching, baseline percentage differences and exact website links. No beam, loads, supports or material E required. Search/get first, then pass only the exact edition-scoped references, never invented property values. Call once for the final shortlist. This compares catalog properties, not member capacity or beam response. No overall winner is implied. For a preliminary calculation, state the assumed inputs before calling the tools and invite corrections after the card. The interactive card already answers the request with drawings, inputs, tables and results. After displaying it, use at most two short sentences for one useful interpretation, tradeoff or missing input. Do not repeat numerical lists, tables, formulas, input summaries, ASCII drawings or another schematic unless the user explicitly asks for them. A short explanation may quote a number when it is needed to explain the observation. For a simple lookup with nothing useful to add, a brief confirmation is enough. In a client without a rendered card, use the data-tool results to provide the requested answer.
