[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7D7ECHQA5BYMECXZ76PBAQZ":3,"skill-same:01M7D7ECHQA5BYMECXZ76PBAQZ":83},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":16,"sourceHtmlUrl":17,"categories":18,"keywords":19,"publishedAt":20,"linkOnly":21,"markdownContent":22,"markdownBody":23,"frontMatter":24,"refName":28,"commitSha":29,"skillDirectory":5,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":32,"ownerHtmlUrl":33,"ownerBlogUrl":34,"ownerDescription":35,"updatedAt":36,"install":37,"resources":50},"01M7D7ECHQA5BYMECXZ76PBAQZ","google-cloud-solution-hybrid-search-alloydb",null,"Discovers requirements and generates architectural, design, and deployment guidance for dynamic hybrid search systems by combining semantic search and keyword search. Optimized for AlloyDB hybrid search use cases in Google Cloud. Use when users need vector search combined with structured SQL filtering, faceted attributes, semantic reranking, in-database AI validation, or serverless hosting across transactional relational databases, analytical data warehouses, or managed database engines. DON'T use this skill for simple keyword-only search, or when a standalone non-relational vector database is required.","tangyixiao\u002Ftangyixiao-skills","tangyixiao","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F115929632?v=4","01M7D582YK6A4M63CAK4CSY4EW","Tang Yixiao","01M7D582Z64DMRSYR872SR34SA",2,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: google-cloud-solution-hybrid-search-alloydb\nmetadata:\n  version: 1.0.1\n  category: MultiProductSolutions\ndescription: Discovers requirements and generates architectural, design, and deployment\n  guidance for dynamic hybrid search systems by combining semantic search and keyword\n  search. Optimized for AlloyDB hybrid search use cases in Google Cloud. Use when\n  users need vector search combined with structured SQL filtering, faceted attributes,\n  semantic reranking, in-database AI validation, or serverless hosting across transactional\n  relational databases, analytical data warehouses, or managed database engines. DON'T\n  use this skill for simple keyword-only search, or when a standalone non-relational\n  vector database is required.\n---\n\n# Dynamic Hybrid Search using AlloyDB\n\nThis skill provides a workflow to design and implement secure, low-latency, and\nhigh-accuracy hybrid search solutions combining structured dataset filtering,\nvector search indexing, faceted metadata filtering, semantic reranking, recall\nevaluation, in-database AI validation, database abstraction layers, and\nserverless application hosting.\n\n## Attribution\n\n**MUST READ**: Before executing any `gcloud`, REST API, or MCP command, you\nMUST read and follow the instructions in\n[Usage Attribution](references\u002Fattribution.md).\n\n## Overview of the workflow\n\nThe workflow consists of the following phases:\n\n1. **Requirements discovery**. Gather detailed requirements related to\n  the cloud workload or use case that the user needs assistance for.\n2. **Solution architecture**. Use the requirements that were gathered\n  in Phase 1 to generate a detailed solution architecture for the cloud\n  workload or use case.\n3. **Solution validation**. Create a plan to validate the generated\n  solution, generate validation instructions and scripts, and run the\n  validation.\n4. **Solution packaging and presentation**. Consolidate the generated\n  content and present the solution.\n\n**Important notes about the workflow**:\n\n- **Strict phase separation**: During Phase 1 (Requirements discovery), when you\n  ask the user clarifying questions, DON'T recommend, propose, or outline any\n  architectural designs, cloud services, or component mappings. This prevents\n  premature architecture commitments or hallucinations before the full scope is\n  understood.\n- **Halting for approval**: For any step where you are instructed\n  to \"obtain approval before proceeding\", you MUST stop executing, present the\n  completed tasks to the user, and wait for their explicit approval. You MUST\n  NOT proceed to execute any subsequent tasks or generate any further guidance\n  in that response.\n- **Ground all generated content**: For all tasks across all phases, you MUST\n  first look in the following resources:\n  - [Product Mapping](references\u002Fproduct-mapping.md),\n  - [Design Recommendations](references\u002Fdesign-recommendations.md) for the\n  required guidance. If the guidance does not provide the required information,\n  you MUST ground the generated content by using the following resources:\n\n  - Google Developer Knowledge MCP server:\n    https:\u002F\u002Fdevelopers.google.com\u002Fknowledge\u002Fmcp.md.txt\n    - Server: https:\u002F\u002Fdeveloperknowledge.googleapis.com\u002Fmcp\n      - Tools:\n        - `developerknowledge:search_documents`\n        - `developerknowledge:get_documents`\n        - `developerknowledge:answer_query`\n  - Relevant skills from https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\n  - Official Google Cloud documentation in\n    [Related Guidance](references\u002Frelated-guidance.md)\n\n## Product Renaming & Terminology\n\nWhen generating solution designs, architecture diagrams, and documentation,\ncheck the latest Google Cloud documentation for the most up-to-date product\nnames. The table below provides examples of name mappings to be aware of. Note\nthat underlying APIs, Terraform resources, and IAM roles may retain their legacy\nidentifiers.\n\n\u003Ctable>\n  \u003Cthead>\n    \u003Ctr>\n      \u003Cth>Legacy Name\u003C\u002Fth>\n      \u003Cth>Updated Name\u003C\u002Fth>\n      \u003Cth>Notes\u003C\u002Fth>\n    \u003C\u002Ftr>\n  \u003C\u002Fthead>\n  \u003Ctbody>\n    \u003Ctr>\n      \u003Ctd>Vertex AI\u003C\u002Ftd>\n      \u003Ctd>Gemini Enterprise Agent Platform\u003C\u002Ftd>\n      \u003Ctd>Gemini Enterprise Agent Platform can be shortened to Agent Platform after first instance\u003C\u002Ftd>\n    \u003C\u002Ftr>\n    \u003Ctr>\n      \u003Ctd>Vertex AI Embedding\u003C\u002Ftd>\n      \u003Ctd>Text embedding on Gemini Enterprise Agent Platform\u003C\u002Ftd>\n      \u003Ctd>This refers to the text embedding models available on Gemini Enterprise Agent Platform\u003C\u002Ftd>\n    \u003C\u002Ftr>\n    \u003Ctr>\n      \u003Ctd>Vertex AI Matching Engine\u003C\u002Ftd>\n      \u003Ctd>Vector Search\u003C\u002Ftd>\n      \u003Ctd>\u003C\u002Ftd>\n    \u003C\u002Ftr>\n  \u003C\u002Ftbody>\n\u003C\u002Ftable>\n\n## Phase 1: Requirements discovery\n\nIn this phase, you must gather detailed requirements related to the hybrid\nsearch workload that the user wants to design and deploy in Google Cloud.\n\n**Acknowledge provided requirements**: If the\nuser's prompt already contains some requirements (functional or non-functional,\nsuch as catalog size, search modalities, faceted attributes, or latency\ntargets), you MUST explicitly acknowledge and restate all of these requirements\nin your response. Do NOT ask the user to describe or re-describe any\nrequirements that they have already provided in the prompt.\n\nComplete the following steps strictly in the specified order:\n\n- [ ] **Step 1**: Ask the user to describe the functional requirements of the\n  workload, including catalog dataset details (e.g., e-commerce apparel, retail\n  products, patent database), search modalities (natural language text, visual\n  search, attribute filters), metadata attributes for faceted filtering (e.g.,\n  `category`, `sub_category`, `color`, `gender`, `price`), and quality checks\n  (reranking, LLM validation).\n- [ ] **Step 2**: You MUST explicitly ask the user to describe ALL of the\n  following six categories of non-functional requirements. You need this\n  information because each category represents a critical architectural pillar,\n  and neglecting any of them can result in a solution that is insecure,\n  unreliable, or inefficient (do NOT omit any of them):\n  - **Security, privacy, and compliance**: E.g., private VPC endpoints, Private\n    Service Connect, Direct VPC Egress, and access control.\n  - **Reliability**: E.g., high availability, failover, disaster recovery goals\n    (RTO\u002FRPO), regional vs multi-region AlloyDB topology.\n  - **Cost**: E.g., budget constraints for compute, database instances, and\n    Gemini Enterprise Agent Platform API calls.\n  - **Operational excellence**: E.g., monitoring, logging, dashboards, and\n    automated deployment.\n  - **Performance**: E.g., target P95 query latency (e.g., \u003C 100ms), vector\n    search recall target (e.g., > 95%), catalog item scale, and QPS\n    expectations.\n  - **Sustainability**: E.g., carbon footprint, low-carbon region selection.\n- [ ] **Step 3**: Ask the user whether the workload currently runs on other\n  cloud providers or on-premises.\n  - If the user's answer is \"yes\", then ask the user to describe the\n    architecture of the current deployment.\n  - If the user's answer is \"no\", then proceed to the next step.\n- [ ] **Step 4**: Ask the user to describe dependencies, if any, on other\n  workloads, products, or tools (e.g., existing inventory databases, ERP\n  systems, application runtime languages like Java or Python).\n\n- [ ] **Step 5**: Review the input that the user has provided so far, and check\n  whether there are any ambiguities, conflicts, or contradictions in the\n  functional requirements, non-functional requirements, and dependencies. You\n  MUST compare all requirements against each other to identify any conflicts.\n\n  If you identify any ambiguities, conflicts, or contradictions in the\n  requirements that the user has provided, you MUST do the following for each\n  ambiguity, conflict, or contradiction:\n  - [ ] Identify exactly where each contradiction lies and explain to the\n    user why the requirements are incompatible and cannot be simultaneously\n    satisfied. Do NOT treat fundamental contradictions as design choice\n    questions (e.g., asking how to implement or configure a conflicting\n    requirement).\n  - [ ] Ask the user to clarify their trade-off preferences to resolve the\n    contradiction.\n  - [ ] If the user delegates the choice to you (e.g., the user replies with\n    \"do what you think is best\" or \"you decide\"), then provide a clear\n    suggestion to resolve the ambiguity or contradiction, explain your\n    reasoning, and ask the user to approve your suggestion.\n\n  **Critical**: Until all the ambiguities and contradictions that you identify\n  are resolved according to the preceding guidance, you must NOT recommend or\n  generate any architecture design or Google Cloud product recommendations.\n\n- [ ] **Step 6**: Summarize the functional and non-functional requirements\n  provided by the user into a consolidated requirements summary.\n\n- [ ] **Step 7**: Present the generated requirements summary to the user and\n  obtain approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Phase 2.\n\n**Important**: **STOP**, DON'T proceed to generate architecture diagram,\narchitecture description or product recommendations until you have confirmed the\ngenerated requirements summary and resolved all ambiguities and contradictions\nin this phase.\n\n## Phase 2: Solution architecture\n\n### Task 2.1: Identify Google Cloud products and features required for the workload.\n\n- [ ] **Step 1**: Recommend products and features that are appropriate for each\n  component of the user's workload, prioritizing Google Cloud products.\n\n  **Important**: The Google Cloud products and features that you recommend\n  MUST be consistent with the guidance in\n  [Product Mapping](references\u002Fproduct-mapping.md).\n\n- [ ] **Step 4**: Present the generated product recommendations to the user and\n  obtain approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Task 2.2.\n\n  **Important**: **STOP**, DON'T proceed to generate architecture diagram until\n  you have confirmed the generated product recommendations with the user.\n\n### Task 2.2: Generate an architecture diagram and description\n\n- [ ] **Step 1**: Generate an architecture diagram in the Mermaid format:\n      https:\u002F\u002Fgithub.com\u002Fmermaid-js\u002Fmermaid.\n\n  The diagram must show the data flows and request flows across the components\n  of the architecture, based on the gathered requirements and product\n  recommendations. The diagram MUST explicitly show both the ingestion pipeline\n  and serving pipeline.\n\n  The following is an **example** of the data flows and request flows that the\n  architecture diagram should show:\n  - **Ingestion pipeline**: Catalog Data -> AlloyDB Table\n    (`apparels`) -> B-Tree Indexes on Facets -> Text embedding\n    (`text-embedding-005`) -> ScaNN Vector Index.\n  - **Serving pipeline**: User Browser -> Cloud Run Web App -> MCP\n    Toolbox for Databases -> AlloyDB Single-Query Hybrid Search (ScaNN Vector\n    Search + SQL WHERE Filters) -> `ai.rank` Reranker -> Gemini Pro\n    `ai.generate` Quality Validation -> Validated Results -> User Browser.\n\n- [ ] **Step 2**: Generate a description that explains the purpose of each\n  component, the relationships between the components, and the task flow or data\n  flow.\n- [ ] **Step 3**: Present the generated architecture diagram and description\n  to the user and obtain approval (the user MUST explicitly say \"yes\" or \"I\n  approve\") before proceeding to Task 2.3.\n\n  **Important**: **STOP**, DON'T proceed to generate design recommendations\n  until you have confirmed the generated architecture description with the user.\n\n### Task 2.3: Generate design recommendations.\n\n- [ ] **Step 1**: Generate design recommendations and best practices to\n  optimally configure each component in the architecture based on the workload\n  requirements.\n\n  **Important**:\n    - When you generate design recommendations, consider the following:\n      - Functional requirements that were gathered in Phase 1.\n      - Non-functional requirements that were gathered in Phase 1.\n    - Align the generated design recommendations with the recommendations in\n      [Design Recommendations](references\u002Fdesign-recommendations.md).\n    - To generate guidance for the non-functional requirements, use the\n      following skills:\n      - `google-cloud-waf-security`\n      - `google-cloud-waf-reliability`\n      - `google-cloud-waf-cost-optimization`\n      - `google-cloud-waf-operational-excellence`\n      - `google-cloud-waf-performance-optimization`\n      - `google-cloud-waf-sustainability`\n- [ ] **Step 2**: Present the generated recommendations to the user and obtain\n  approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Task 2.4.\n\n  **Important**: **STOP**, DON'T proceed to generate deployment guidance until\n  you have confirmed the design recommendations with the user.\n\n### Task 2.4: Generate deployment guidance.\n\n- [ ] **Step 1**: Generate guidance to deploy the solution, including the\n  following:\n  - AlloyDB DDL & SQL setup scripts for extensions (`google_ml_integration`,\n    `alloydb_scan`), tables, B-Tree indexes, ScaNN vector indexes, hybrid search\n    SQL, and Gemini validation CTEs.\n  - MCP Toolbox deployment configuration on Cloud Run.\n  - Python Cloud Run Function shim deployment command.\n  - Application deployment command (`gcloud run deploy {app_name}`).\n  - Terraform code or `gcloud` CLI commands to create required infrastructure.\n\n  **Important**: The deployment guidance that you generate MUST be consistent\n  with the guidance in the following resources:\n  - [Related Guidance](references\u002Frelated-guidance.md)\n  - Relevant skills in\n    https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Ftree\u002Fmain\u002Fskills\u002Fcloud\n\n- [ ] **Step 2**: Present the generated deployment guidance to the user and\n  obtain approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Phase 3.\n\n  **Important**: **STOP**, DON'T proceed to generate solution validation until\n  you have confirmed the deployment guidance with the user.\n\n## Phase 3: Solution validation\n\n### Task 3.1: Pre-deployment validation\n\n- [ ] **Step 1**: Create a pre-deployment plan to statically validate the\n  generated solution and verify that it meets the workload requirements\n  without provisioning live resources:\n  - **Deployment dry-run**: Validate infrastructure syntax and preview the\n    resources that will be provisioned using dry-run commands (e.g.,\n    `terraform plan` or (where supported) `gcloud ... --dry-run`).\n  - **Architecture & policy analysis**: Perform static verification of\n    network routing topologies, firewall rules, and IAM enforcement against\n    best practices.\n- [ ] **Step 2**: Present the static validation plan to the user, obtain\n  approval (the user MUST explicitly say \"yes\" or \"I approve\"), and execute the\n  dry-run commands.\n- [ ] **Step 3**: Troubleshoot and fix any errors or policy discrepancies\n  identified during dry-run checks until validation succeeds.\n- [ ] **Step 4**: Proceed to Task 3.2\n\n### Task 3.2: Runtime validation (Post-deployment)\n\n- [ ] **Step 1**: Ask the user whether they choose to deploy the infrastructure\n  now to perform live runtime verification, or skip directly to Phase 4.\n- [ ] **Step 2**: **If the user chooses to deploy the infrastructure**:\n  - After the user deploys the infrastructure, generate runtime\n    verification commands (using tools like `curl`, `ping`, or `gcloud`)\n    and provide them to the user to execute, to test live endpoint\n    reachability, networking paths, and load balancer routing.\n  - Troubleshoot any deployment or runtime routing issues until checks pass.\n- [ ] **Step 3**: Proceed to Phase 4.\n\n## Phase 4: Solution packaging and presentation\n\n- [ ] **Step 1**: Consolidate the final text artifacts that were generated in\n  Phase 2 into a single Markdown file named `solution-architecture-guide.md`,\n  based on the template in [Output Template](assets\u002Foutput-template.md).\n- [ ] **Step 2**: Request the user's permission to write the code files in the\n  user's workspace.\n- [ ] **Step 3**: After the user gives permission, write the final code files in\n  the user's workspace.\n","# Dynamic Hybrid Search using AlloyDB\n\nThis skill provides a workflow to design and implement secure, low-latency, and\nhigh-accuracy hybrid search solutions combining structured dataset filtering,\nvector search indexing, faceted metadata filtering, semantic reranking, recall\nevaluation, in-database AI validation, database abstraction layers, and\nserverless application hosting.\n\n## Attribution\n\n**MUST READ**: Before executing any `gcloud`, REST API, or MCP command, you\nMUST read and follow the instructions in\n[Usage Attribution](references\u002Fattribution.md).\n\n## Overview of the workflow\n\nThe workflow consists of the following phases:\n\n1. **Requirements discovery**. Gather detailed requirements related to\n  the cloud workload or use case that the user needs assistance for.\n2. **Solution architecture**. Use the requirements that were gathered\n  in Phase 1 to generate a detailed solution architecture for the cloud\n  workload or use case.\n3. **Solution validation**. Create a plan to validate the generated\n  solution, generate validation instructions and scripts, and run the\n  validation.\n4. **Solution packaging and presentation**. Consolidate the generated\n  content and present the solution.\n\n**Important notes about the workflow**:\n\n- **Strict phase separation**: During Phase 1 (Requirements discovery), when you\n  ask the user clarifying questions, DON'T recommend, propose, or outline any\n  architectural designs, cloud services, or component mappings. This prevents\n  premature architecture commitments or hallucinations before the full scope is\n  understood.\n- **Halting for approval**: For any step where you are instructed\n  to \"obtain approval before proceeding\", you MUST stop executing, present the\n  completed tasks to the user, and wait for their explicit approval. You MUST\n  NOT proceed to execute any subsequent tasks or generate any further guidance\n  in that response.\n- **Ground all generated content**: For all tasks across all phases, you MUST\n  first look in the following resources:\n  - [Product Mapping](references\u002Fproduct-mapping.md),\n  - [Design Recommendations](references\u002Fdesign-recommendations.md) for the\n  required guidance. If the guidance does not provide the required information,\n  you MUST ground the generated content by using the following resources:\n\n  - Google Developer Knowledge MCP server:\n    https:\u002F\u002Fdevelopers.google.com\u002Fknowledge\u002Fmcp.md.txt\n    - Server: https:\u002F\u002Fdeveloperknowledge.googleapis.com\u002Fmcp\n      - Tools:\n        - `developerknowledge:search_documents`\n        - `developerknowledge:get_documents`\n        - `developerknowledge:answer_query`\n  - Relevant skills from https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\n  - Official Google Cloud documentation in\n    [Related Guidance](references\u002Frelated-guidance.md)\n\n## Product Renaming & Terminology\n\nWhen generating solution designs, architecture diagrams, and documentation,\ncheck the latest Google Cloud documentation for the most up-to-date product\nnames. The table below provides examples of name mappings to be aware of. Note\nthat underlying APIs, Terraform resources, and IAM roles may retain their legacy\nidentifiers.\n\n\u003Ctable>\n  \u003Cthead>\n    \u003Ctr>\n      \u003Cth>Legacy Name\u003C\u002Fth>\n      \u003Cth>Updated Name\u003C\u002Fth>\n      \u003Cth>Notes\u003C\u002Fth>\n    \u003C\u002Ftr>\n  \u003C\u002Fthead>\n  \u003Ctbody>\n    \u003Ctr>\n      \u003Ctd>Vertex AI\u003C\u002Ftd>\n      \u003Ctd>Gemini Enterprise Agent Platform\u003C\u002Ftd>\n      \u003Ctd>Gemini Enterprise Agent Platform can be shortened to Agent Platform after first instance\u003C\u002Ftd>\n    \u003C\u002Ftr>\n    \u003Ctr>\n      \u003Ctd>Vertex AI Embedding\u003C\u002Ftd>\n      \u003Ctd>Text embedding on Gemini Enterprise Agent Platform\u003C\u002Ftd>\n      \u003Ctd>This refers to the text embedding models available on Gemini Enterprise Agent Platform\u003C\u002Ftd>\n    \u003C\u002Ftr>\n    \u003Ctr>\n      \u003Ctd>Vertex AI Matching Engine\u003C\u002Ftd>\n      \u003Ctd>Vector Search\u003C\u002Ftd>\n      \u003Ctd>\u003C\u002Ftd>\n    \u003C\u002Ftr>\n  \u003C\u002Ftbody>\n\u003C\u002Ftable>\n\n## Phase 1: Requirements discovery\n\nIn this phase, you must gather detailed requirements related to the hybrid\nsearch workload that the user wants to design and deploy in Google Cloud.\n\n**Acknowledge provided requirements**: If the\nuser's prompt already contains some requirements (functional or non-functional,\nsuch as catalog size, search modalities, faceted attributes, or latency\ntargets), you MUST explicitly acknowledge and restate all of these requirements\nin your response. Do NOT ask the user to describe or re-describe any\nrequirements that they have already provided in the prompt.\n\nComplete the following steps strictly in the specified order:\n\n- [ ] **Step 1**: Ask the user to describe the functional requirements of the\n  workload, including catalog dataset details (e.g., e-commerce apparel, retail\n  products, patent database), search modalities (natural language text, visual\n  search, attribute filters), metadata attributes for faceted filtering (e.g.,\n  `category`, `sub_category`, `color`, `gender`, `price`), and quality checks\n  (reranking, LLM validation).\n- [ ] **Step 2**: You MUST explicitly ask the user to describe ALL of the\n  following six categories of non-functional requirements. You need this\n  information because each category represents a critical architectural pillar,\n  and neglecting any of them can result in a solution that is insecure,\n  unreliable, or inefficient (do NOT omit any of them):\n  - **Security, privacy, and compliance**: E.g., private VPC endpoints, Private\n    Service Connect, Direct VPC Egress, and access control.\n  - **Reliability**: E.g., high availability, failover, disaster recovery goals\n    (RTO\u002FRPO), regional vs multi-region AlloyDB topology.\n  - **Cost**: E.g., budget constraints for compute, database instances, and\n    Gemini Enterprise Agent Platform API calls.\n  - **Operational excellence**: E.g., monitoring, logging, dashboards, and\n    automated deployment.\n  - **Performance**: E.g., target P95 query latency (e.g., \u003C 100ms), vector\n    search recall target (e.g., > 95%), catalog item scale, and QPS\n    expectations.\n  - **Sustainability**: E.g., carbon footprint, low-carbon region selection.\n- [ ] **Step 3**: Ask the user whether the workload currently runs on other\n  cloud providers or on-premises.\n  - If the user's answer is \"yes\", then ask the user to describe the\n    architecture of the current deployment.\n  - If the user's answer is \"no\", then proceed to the next step.\n- [ ] **Step 4**: Ask the user to describe dependencies, if any, on other\n  workloads, products, or tools (e.g., existing inventory databases, ERP\n  systems, application runtime languages like Java or Python).\n\n- [ ] **Step 5**: Review the input that the user has provided so far, and check\n  whether there are any ambiguities, conflicts, or contradictions in the\n  functional requirements, non-functional requirements, and dependencies. You\n  MUST compare all requirements against each other to identify any conflicts.\n\n  If you identify any ambiguities, conflicts, or contradictions in the\n  requirements that the user has provided, you MUST do the following for each\n  ambiguity, conflict, or contradiction:\n  - [ ] Identify exactly where each contradiction lies and explain to the\n    user why the requirements are incompatible and cannot be simultaneously\n    satisfied. Do NOT treat fundamental contradictions as design choice\n    questions (e.g., asking how to implement or configure a conflicting\n    requirement).\n  - [ ] Ask the user to clarify their trade-off preferences to resolve the\n    contradiction.\n  - [ ] If the user delegates the choice to you (e.g., the user replies with\n    \"do what you think is best\" or \"you decide\"), then provide a clear\n    suggestion to resolve the ambiguity or contradiction, explain your\n    reasoning, and ask the user to approve your suggestion.\n\n  **Critical**: Until all the ambiguities and contradictions that you identify\n  are resolved according to the preceding guidance, you must NOT recommend or\n  generate any architecture design or Google Cloud product recommendations.\n\n- [ ] **Step 6**: Summarize the functional and non-functional requirements\n  provided by the user into a consolidated requirements summary.\n\n- [ ] **Step 7**: Present the generated requirements summary to the user and\n  obtain approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Phase 2.\n\n**Important**: **STOP**, DON'T proceed to generate architecture diagram,\narchitecture description or product recommendations until you have confirmed the\ngenerated requirements summary and resolved all ambiguities and contradictions\nin this phase.\n\n## Phase 2: Solution architecture\n\n### Task 2.1: Identify Google Cloud products and features required for the workload.\n\n- [ ] **Step 1**: Recommend products and features that are appropriate for each\n  component of the user's workload, prioritizing Google Cloud products.\n\n  **Important**: The Google Cloud products and features that you recommend\n  MUST be consistent with the guidance in\n  [Product Mapping](references\u002Fproduct-mapping.md).\n\n- [ ] **Step 4**: Present the generated product recommendations to the user and\n  obtain approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Task 2.2.\n\n  **Important**: **STOP**, DON'T proceed to generate architecture diagram until\n  you have confirmed the generated product recommendations with the user.\n\n### Task 2.2: Generate an architecture diagram and description\n\n- [ ] **Step 1**: Generate an architecture diagram in the Mermaid format:\n      https:\u002F\u002Fgithub.com\u002Fmermaid-js\u002Fmermaid.\n\n  The diagram must show the data flows and request flows across the components\n  of the architecture, based on the gathered requirements and product\n  recommendations. The diagram MUST explicitly show both the ingestion pipeline\n  and serving pipeline.\n\n  The following is an **example** of the data flows and request flows that the\n  architecture diagram should show:\n  - **Ingestion pipeline**: Catalog Data -> AlloyDB Table\n    (`apparels`) -> B-Tree Indexes on Facets -> Text embedding\n    (`text-embedding-005`) -> ScaNN Vector Index.\n  - **Serving pipeline**: User Browser -> Cloud Run Web App -> MCP\n    Toolbox for Databases -> AlloyDB Single-Query Hybrid Search (ScaNN Vector\n    Search + SQL WHERE Filters) -> `ai.rank` Reranker -> Gemini Pro\n    `ai.generate` Quality Validation -> Validated Results -> User Browser.\n\n- [ ] **Step 2**: Generate a description that explains the purpose of each\n  component, the relationships between the components, and the task flow or data\n  flow.\n- [ ] **Step 3**: Present the generated architecture diagram and description\n  to the user and obtain approval (the user MUST explicitly say \"yes\" or \"I\n  approve\") before proceeding to Task 2.3.\n\n  **Important**: **STOP**, DON'T proceed to generate design recommendations\n  until you have confirmed the generated architecture description with the user.\n\n### Task 2.3: Generate design recommendations.\n\n- [ ] **Step 1**: Generate design recommendations and best practices to\n  optimally configure each component in the architecture based on the workload\n  requirements.\n\n  **Important**:\n    - When you generate design recommendations, consider the following:\n      - Functional requirements that were gathered in Phase 1.\n      - Non-functional requirements that were gathered in Phase 1.\n    - Align the generated design recommendations with the recommendations in\n      [Design Recommendations](references\u002Fdesign-recommendations.md).\n    - To generate guidance for the non-functional requirements, use the\n      following skills:\n      - `google-cloud-waf-security`\n      - `google-cloud-waf-reliability`\n      - `google-cloud-waf-cost-optimization`\n      - `google-cloud-waf-operational-excellence`\n      - `google-cloud-waf-performance-optimization`\n      - `google-cloud-waf-sustainability`\n- [ ] **Step 2**: Present the generated recommendations to the user and obtain\n  approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Task 2.4.\n\n  **Important**: **STOP**, DON'T proceed to generate deployment guidance until\n  you have confirmed the design recommendations with the user.\n\n### Task 2.4: Generate deployment guidance.\n\n- [ ] **Step 1**: Generate guidance to deploy the solution, including the\n  following:\n  - AlloyDB DDL & SQL setup scripts for extensions (`google_ml_integration`,\n    `alloydb_scan`), tables, B-Tree indexes, ScaNN vector indexes, hybrid search\n    SQL, and Gemini validation CTEs.\n  - MCP Toolbox deployment configuration on Cloud Run.\n  - Python Cloud Run Function shim deployment command.\n  - Application deployment command (`gcloud run deploy {app_name}`).\n  - Terraform code or `gcloud` CLI commands to create required infrastructure.\n\n  **Important**: The deployment guidance that you generate MUST be consistent\n  with the guidance in the following resources:\n  - [Related Guidance](references\u002Frelated-guidance.md)\n  - Relevant skills in\n    https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Ftree\u002Fmain\u002Fskills\u002Fcloud\n\n- [ ] **Step 2**: Present the generated deployment guidance to the user and\n  obtain approval (the user MUST explicitly say \"yes\" or \"I approve\") before\n  proceeding to Phase 3.\n\n  **Important**: **STOP**, DON'T proceed to generate solution validation until\n  you have confirmed the deployment guidance with the user.\n\n## Phase 3: Solution validation\n\n### Task 3.1: Pre-deployment validation\n\n- [ ] **Step 1**: Create a pre-deployment plan to statically validate the\n  generated solution and verify that it meets the workload requirements\n  without provisioning live resources:\n  - **Deployment dry-run**: Validate infrastructure syntax and preview the\n    resources that will be provisioned using dry-run commands (e.g.,\n    `terraform plan` or (where supported) `gcloud ... --dry-run`).\n  - **Architecture & policy analysis**: Perform static verification of\n    network routing topologies, firewall rules, and IAM enforcement against\n    best practices.\n- [ ] **Step 2**: Present the static validation plan to the user, obtain\n  approval (the user MUST explicitly say \"yes\" or \"I approve\"), and execute the\n  dry-run commands.\n- [ ] **Step 3**: Troubleshoot and fix any errors or policy discrepancies\n  identified during dry-run checks until validation succeeds.\n- [ ] **Step 4**: Proceed to Task 3.2\n\n### Task 3.2: Runtime validation (Post-deployment)\n\n- [ ] **Step 1**: Ask the user whether they choose to deploy the infrastructure\n  now to perform live runtime verification, or skip directly to Phase 4.\n- [ ] **Step 2**: **If the user chooses to deploy the infrastructure**:\n  - After the user deploys the infrastructure, generate runtime\n    verification commands (using tools like `curl`, `ping`, or `gcloud`)\n    and provide them to the user to execute, to test live endpoint\n    reachability, networking paths, and load balancer routing.\n  - Troubleshoot any deployment or runtime routing issues until checks pass.\n- [ ] **Step 3**: Proceed to Phase 4.\n\n## Phase 4: Solution packaging and presentation\n\n- [ ] **Step 1**: Consolidate the final text artifacts that were generated in\n  Phase 2 into a single Markdown file named `solution-architecture-guide.md`,\n  based on the template in [Output Template](assets\u002Foutput-template.md).\n- [ ] **Step 2**: Request the user's permission to write the code files in the\n  user's workspace.\n- [ ] **Step 3**: After the user gives permission, write the final code files in\n  the user's workspace.\n",{"name":5,"metadata":25,"description":7},{"version":26,"category":27},"1.0.1","MultiProductSolutions","main","2a870f189175a48c39f5f7e26a6d93b0be9cd13b",0,"https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills","https:\u002F\u002Ftangyixiao.github.io\u002Ftangyixiao-skills\u002F","https:\u002F\u002Fgithub.com\u002Ftangyixiao","http:\u002F\u002Fwww.sxyz.net\u002F","I am Tang Yixiao, a student in Shaoxing No.1 High School. I am eager to learn on GitHub and contribute my part to this community.","2026-11-11 15:45:59",{"source":8,"skillName":5,"commands":38,"prompt":48,"zipAvailable":49},[39,42,45],{"tool":40,"command":41},"npx","npx skills add tangyixiao\u002Ftangyixiao-skills --skill google-cloud-solution-hybrid-search-alloydb",{"tool":43,"command":44},"bunx","bunx skills add tangyixiao\u002Ftangyixiao-skills --skill google-cloud-solution-hybrid-search-alloydb",{"tool":46,"command":47},"pnpm","pnpm dlx skills add tangyixiao\u002Ftangyixiao-skills --skill google-cloud-solution-hybrid-search-alloydb","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002FSKILL.md",true,[51,57,63,67,71,75,79],{"path":52,"resourceType":53,"resourceTypeDesc":54,"sizeBytes":55,"sourceHtmlUrl":56},"assets\u002Foutput-template.md","ASSET","资产文件",7345,"https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002Fassets\u002Foutput-template.md",{"path":58,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":61,"sourceHtmlUrl":62},"INSTALL-SOURCE.json","REFERENCE","参考文档",589,"https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002FINSTALL-SOURCE.json",{"path":64,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":66,"sourceHtmlUrl":16},"LICENSE","许可证",11358,{"path":68,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":69,"sourceHtmlUrl":70},"references\u002Fattribution.md",1977,"https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002Freferences\u002Fattribution.md",{"path":72,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":73,"sourceHtmlUrl":74},"references\u002Fdesign-recommendations.md",5992,"https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002Freferences\u002Fdesign-recommendations.md",{"path":76,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":77,"sourceHtmlUrl":78},"references\u002Fproduct-mapping.md",10957,"https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002Freferences\u002Fproduct-mapping.md",{"path":80,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":81,"sourceHtmlUrl":82},"references\u002Frelated-guidance.md",2617,"https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fgoogle-cloud-solution-hybrid-search-alloydb\u002Freferences\u002Frelated-guidance.md",[84,92,101,109,117,125],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":88,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":20},"01M7D8JK8FYBFNNAJXWD89CFH7","xlsx","Use this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like \"the xlsx in my downloads\") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fxlsx\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fxlsx\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":96,"licenseFileUrl":97,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":20},"01M7D8JG8ERGGQQAMX0DWCG9GR","writing-shape","Writing, exploit: shape raw material into an article, paragraph by paragraph.","MIT","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-shape\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-shape\u002FSKILL.md",[],[],{"skillNo":102,"name":103,"nameZh":6,"description":104,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":96,"licenseFileUrl":105,"sourceHtmlUrl":106,"categories":107,"keywords":108,"publishedAt":20},"01M7D8JE672GW3EHSE8EF185M4","writing-fragments","Writing, explore: mine raw fragments, no structure yet.","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-fragments\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-fragments\u002FSKILL.md",[],[],{"skillNo":110,"name":111,"nameZh":6,"description":112,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":96,"licenseFileUrl":113,"sourceHtmlUrl":114,"categories":115,"keywords":116,"publishedAt":20},"01M7D8JBK572YDFMND8Q6CEYJW","writing-for-agents","Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-for-agents\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-for-agents\u002FSKILL.md",[],[],{"skillNo":118,"name":119,"nameZh":6,"description":120,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":96,"licenseFileUrl":121,"sourceHtmlUrl":122,"categories":123,"keywords":124,"publishedAt":20},"01M7D8J5QPWEXX61EGBKYH73R8","writing-clearly-and-concisely","Use when writing prose humans will read—documentation, commit messages, error messages, explanations, reports, or UI text. Applies Strunk's timeless rules for clearer, stronger, more professional writing.","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-clearly-and-concisely\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-clearly-and-concisely\u002FSKILL.md",[],[],{"skillNo":126,"name":127,"nameZh":6,"description":128,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":96,"licenseFileUrl":129,"sourceHtmlUrl":130,"categories":131,"keywords":132,"publishedAt":20},"01M7D8J36YWC8WZS5DK5AH03CD","writing-beats","Writing, exploit; assemble raw material into a journey of beats, grounding each term before a beat leans on it.","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-beats\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Ftangyixiao\u002Ftangyixiao-skills\u002Fblob\u002Fmain\u002Fwriting-beats\u002FSKILL.md",[],[]]