Patsnap Solution Engine

This MCP generates TRIZ concept solutions for R&D problem solving and product cost reduction. Based on the task objective, it enters either an innovation workflow or a cost-reduction workflow: the innovation workflow uses problem analysis, case matching, and TRIZ methods to generate concept solutions; the cost-reduction workflow uses product structure, function, and cost information with TRIZ/DFMA diagnosis and trimming to generate cost-reduction concepts and benefit estimates. It supports task progress tracking, candidate retrieval, and solution-detail lookup so enterprise agents can efficiently produce evaluable concept solutions. It provides 6 tools across two capability areas. TRIZ innovation includes run_triz_innovation_task, fetch_triz_innovation_task_stream, and fetch_triz_innovation_solution_detail; TRIZ cost reduction includes run_triz_reduction_task, fetch_triz_reduction_task_stream, and fetch_triz_reduction_solution_detail. The standard call order is: use task_run to create and submit the background task, use task_stream to retrieve progress and candidate solutions, and then use solution_detail when a specific candidate needs to be expanded. A complete solution-generation run usually takes about 5-10 minutes; actual time may vary with task complexity, input completeness, and system load, so clients should reserve sufficient processing time.

• When generating TRIZ innovation concepts around technical contradictions, performance targets, structural design, or system constraints • When exploring cost reduction through part consolidation, material substitution, manufacturing processes, assembly steps, or sourcing constraints • When an R&D agent needs progress tracking and multiple review-ready candidate concepts • When expanding one selected candidate into a more detailed solution after task_stream returns candidates • Not as a substitute for final engineering validation, production approval, or a verified cost commitment

• run_triz_innovation_task requires user_input; include the engineering problem, desired outcome, current approach, and important constraints when possible. The limit is 1,000 characters in the CN region and 500 characters in other regions • fetch_triz_innovation_task_stream requires the job_id returned by the innovation task_run tool to retrieve progress, messages, and candidate concepts • fetch_triz_innovation_solution_detail requires job_id and candidate idea_id to expand a selected innovation concept • run_triz_reduction_task requires user_input; describe the product or component, cost drivers, current material/process/assembly design, and non-negotiable performance constraints. The limit is 1,000 characters in the CN region and 500 characters in other regions • fetch_triz_reduction_task_stream requires the job_id returned by the cost-reduction task_run tool to retrieve progress, messages, and candidate concepts • fetch_triz_reduction_solution_detail requires job_id and candidate idea_id to expand a selected cost-reduction concept • Use IDs from the same workflow branch; internal tokens, providers, and gateway credentials are not Tool parameters

• task_run returns the task-submission result, typically including job_id, task_id, branch, created, auto_mode, status, and the recommended next_tool • task_stream retrieves progress, tool outputs, assistant messages, and candidate concepts. Its final result typically includes job_id, task_id, status, terminal_event, last_event, candidate ideas, candidate count, recommended idea ID/title, final message, display Markdown, solution images, and other available content • TRIZ innovation stream results may include structured modules such as system_component_analysis, component_touch_analysis, functional_modeling, system_structure_analysis, and causal_chain_analysis • TRIZ cost-reduction stream results may include structured modules such as component_cost_mapping, cost_component, cost_trim, dfma_component, dfma_directions, and dfma_concept_solutions • solution_detail returns solution_detail together with the related job_id, task_id, idea_id, status, and available display content • Clients should retain job_id and candidate idea_id, and handle empty candidates, optional fields, and non-success workflow states

• Current CN production endpoint: https://ai-fabric.zhihuiya.com/mcp/patsnap-solution-engine?APP_ID=Patsnap • 6 tools are available: 3 for TRIZ innovation and 3 for TRIZ cost reduction • The standard call chain is task_run -> task_stream -> solution_detail; solution_detail is needed only when expanding a selected candidate • task_run creates and submits the background task; task_stream retrieves progress and results, and does not start the workflow • run_triz_innovation_task and run_triz_reduction_task use credits_zh=1800 for Chinese and credits_en=4500 for English • fetch_triz_innovation_task_stream and fetch_triz_reduction_task_stream use credits_zh=0 and credits_en=0 • fetch_triz_innovation_solution_detail and fetch_triz_reduction_solution_detail use credits_zh=200 for Chinese and credits_en=500 for English • The user_input limit for both task_run tools is 1,000 characters in the CN region and 500 characters in other regions • A complete solution-generation run usually takes about 5-10 minutes; actual time may vary with task complexity, input completeness, and system load, so clients should reserve sufficient processing time • Tool response acceptance: overall p95 <= 10 seconds. Current records show innovation task initialization p95 at 4,292 ms, cost-reduction task initialization p95 at 2,765 ms, and overall p95 at 4,292 ms, so the result passed • Concurrency acceptance: 10-concurrency failure rate < 5%; both capability areas recorded 0.00% and passed • Functional validity is 40/40 (100.00%); performance metrics cover the two task_run initialization tools, while task_stream and solution_detail are not included in that performance statistic • Generated content expands the solution space and still requires validation against business data and engineering review for feasibility, innovation value, and realized cost reduction