generated: '2026-09-05' method: derived source: >- mcp/3vjia-technology-mcp.yml (candidate tools) bound to openapi/3vjia-technology-open-platform-openapi.yml operationIds note: >- 3vjia ships NO MCP server and NO GraphQL surface, so this is not a crosswalk between two shipped interfaces. It is the forward map from the candidate tool set to the REST operations that would back it, plus — the part that is actually load-bearing — the record of how much of the 429-operation REST surface no plausible tool set reaches. Confidence is `high` only where a documented operationId is named; the candidate tools were bound by name and semantics against the derived spec. surfaces: openapi: file: openapi/3vjia-technology-open-platform-openapi.yml operations: 429 gated: true note: enterprise OAuth 2.0 client_credentials; anonymous calls return code 100100002 graphql: endpoint: null present: false mcp: url: null present: false crosswalk: - tool: list_design_schemes category: schemes rest: [apiV1SchemeGetSchemeListByPage, apiV2SchemeGetSchemeListByPage] binding: one-to-one confidence: high note: v1 and v2 of the same operation are both live and both documented; a tool has to choose. - tool: get_design_scheme category: schemes rest: [apiV1LayoutSchemeGetDetail] binding: one-to-one confidence: medium note: >- several scheme-detail operations exist across product lines (apiV1LayoutSchemeGetDetail for layout schemes, apiV1SchemesLandGetGoodsDetail for Design Island listings); the right one depends on which product the tenant licenses. - tool: get_scheme_product_list category: schemes rest: [apiV1SchemeGetSchemeProduct, apiV1QuotationGetSchemeProductPrice] binding: composite confidence: high note: 获取方案产品清单接口 and 获取方案成品清单接口 are separate operations covering the raw and priced BOM. - tool: start_quotation category: quotation rest: [apiV1QuotationQuoteScheme, apiV1QuotationQuoteOrder] binding: one-to-many confidence: high note: two variants — scheme-level (quoteScheme) and order-level (quoteOrder). - tool: get_quotation_result category: quotation rest: [apiV1QuotationQueryQuoteSchemeResult, apiV1QuotationQueryQuoteOrderResult] binding: one-to-many confidence: high note: poll the variant matching the submit call. Terminal condition is errorCode != 609006. - tool: return_factory_order category: orders rest: [apiV1AimesFactoryOrderOpReturnFactoryOrder] binding: one-to-one confidence: high note: must be preceded by apiV1AimesFactoryOrderOpGetReturnSetting to learn which nodes still allow a return. - tool: list_accounts category: accounts rest: [apiV1UserListByPage, apiV1DepartmentGetDeptListByPage, apiV1DepartmentGetDeptTreeList] binding: composite confidence: high - tool: bind_account category: accounts rest: [apiV1AccountbindingBindAccount, apiV1AccountbindingUnBindAccount, apiV1AccountbindingGetAccountBindingRelation] binding: inverse-pair confidence: high mcp_only: [] rest_only: count_estimate: ~400 note: >- The overwhelming majority of the 429 operations have no plausible agent tool, and the reason is structural rather than an oversight on our part. Three groups: groups: - group: 专属服务 (dedicated services) operations: 17 sub-categories keyed to individual enterprise ids (C00000022, C00000868, C11876880, ...) reason: >- single-customer operations. They are published in the public catalog but are only callable by the one tenant they were built for, so no general tool can wrap them. - group: AIMES / MOS shop-floor operations reason: >- packaging, batching, scheduling optimisation, material management and production registration — machine-to-machine MES traffic with no agent-shaped intent. - group: fine-grained CRUD reason: >- dozens of near-identical add/edit/delete operations per entity (materials, categories, departments, shops, channels) that a tool set would collapse into a handful of parameterised tools. coverage: rest_operations: 429 tools_defined: 18 bound_with_confirmed_operation_ids: 8 mcp_only: 0 graphql_only: 0