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.
Permissions
Files
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.
This skill provides a workflow to design and implement secure, low-latency, and high-accuracy hybrid search solutions combining structured dataset filtering, vector search indexing, faceted metadata filtering, semantic reranking, recall evaluation, in-database AI validation, database abstraction layers, and serverless application hosting.
Version history
The workflow consists of the following phases:
Important notes about the workflow:
Product Mapping,
Design Recommendations for the required guidance. If the guidance does not provide the required information, you MUST ground the generated content by using the following resources:
Google Developer Knowledge MCP server: https://developers.google.com/knowledge/mcp.md.txt
developerknowledge:search_documentsdeveloperknowledge:get_documentsdeveloperknowledge:answer_queryRelevant skills from https://github.com/google/skills
Official Google Cloud documentation in Related Guidance
When generating solution designs, architecture diagrams, and documentation, check the latest Google Cloud documentation for the most up-to-date product names. The table below provides examples of name mappings to be aware of. Note that underlying APIs, Terraform resources, and IAM roles may retain their legacy identifiers.
In this phase, you must gather detailed requirements related to the hybrid search workload that the user wants to design and deploy in Google Cloud.
Acknowledge provided requirements: If the user's prompt already contains some requirements (functional or non-functional, such as catalog size, search modalities, faceted attributes, or latency targets), you MUST explicitly acknowledge and restate all of these requirements in your response. Do NOT ask the user to describe or re-describe any requirements that they have already provided in the prompt.
Complete the following steps strictly in the specified order:
Step 1: Ask the user to describe the functional requirements of the
workload, including catalog dataset details (e.g., e-commerce apparel, retail
products, patent database), search modalities (natural language text, visual
search, attribute filters), metadata attributes for faceted filtering (e.g.,
category, sub_category, color, gender, price), and quality checks
(reranking, LLM validation).
Step 2: You MUST explicitly ask the user to describe ALL of the following six categories of non-functional requirements. You need this information because each category represents a critical architectural pillar, and neglecting any of them can result in a solution that is insecure, unreliable, or inefficient (do NOT omit any of them):
Step 3: Ask the user whether the workload currently runs on other cloud providers or on-premises.
Step 4: Ask the user to describe dependencies, if any, on other workloads, products, or tools (e.g., existing inventory databases, ERP systems, application runtime languages like Java or Python).
Step 5: Review the input that the user has provided so far, and check whether there are any ambiguities, conflicts, or contradictions in the functional requirements, non-functional requirements, and dependencies. You MUST compare all requirements against each other to identify any conflicts.
If you identify any ambiguities, conflicts, or contradictions in the requirements that the user has provided, you MUST do the following for each ambiguity, conflict, or contradiction:
Critical: Until all the ambiguities and contradictions that you identify are resolved according to the preceding guidance, you must NOT recommend or generate any architecture design or Google Cloud product recommendations.
Step 6: Summarize the functional and non-functional requirements provided by the user into a consolidated requirements summary.
Step 7: Present the generated requirements summary to the user and obtain approval (the user MUST explicitly say "yes" or "I approve") before proceeding to Phase 2.
Important: STOP, DON'T proceed to generate architecture diagram, architecture description or product recommendations until you have confirmed the generated requirements summary and resolved all ambiguities and contradictions in this phase.
Step 1: Recommend products and features that are appropriate for each component of the user's workload, prioritizing Google Cloud products.
Important: The Google Cloud products and features that you recommend MUST be consistent with the guidance in Product Mapping.
Step 4: Present the generated product recommendations to the user and obtain approval (the user MUST explicitly say "yes" or "I approve") before proceeding to Task 2.2.
Important: STOP, DON'T proceed to generate architecture diagram until you have confirmed the generated product recommendations with the user.
Step 1: Generate an architecture diagram in the Mermaid format: https://github.com/mermaid-js/mermaid.
The diagram must show the data flows and request flows across the components of the architecture, based on the gathered requirements and product recommendations. The diagram MUST explicitly show both the ingestion pipeline and serving pipeline.
The following is an example of the data flows and request flows that the architecture diagram should show:
apparels) -> B-Tree Indexes on Facets -> Text embedding
(text-embedding-005) -> ScaNN Vector Index.ai.rank Reranker -> Gemini Pro
ai.generate Quality Validation -> Validated Results -> User Browser.Step 2: Generate a description that explains the purpose of each component, the relationships between the components, and the task flow or data flow.
Step 3: Present the generated architecture diagram and description to the user and obtain approval (the user MUST explicitly say "yes" or "I approve") before proceeding to Task 2.3.
Important: STOP, DON'T proceed to generate design recommendations until you have confirmed the generated architecture description with the user.
Step 1: Generate design recommendations and best practices to optimally configure each component in the architecture based on the workload requirements.
Important:
google-cloud-waf-securitygoogle-cloud-waf-reliabilitygoogle-cloud-waf-cost-optimizationgoogle-cloud-waf-operational-excellencegoogle-cloud-waf-performance-optimizationgoogle-cloud-waf-sustainabilityStep 2: Present the generated recommendations to the user and obtain approval (the user MUST explicitly say "yes" or "I approve") before proceeding to Task 2.4.
Important: STOP, DON'T proceed to generate deployment guidance until you have confirmed the design recommendations with the user.
Step 1: Generate guidance to deploy the solution, including the following:
google_ml_integration,
alloydb_scan), tables, B-Tree indexes, ScaNN vector indexes, hybrid search
SQL, and Gemini validation CTEs.gcloud run deploy {app_name}).gcloud CLI commands to create required infrastructure.Important: The deployment guidance that you generate MUST be consistent with the guidance in the following resources:
Step 2: Present the generated deployment guidance to the user and obtain approval (the user MUST explicitly say "yes" or "I approve") before proceeding to Phase 3.
Important: STOP, DON'T proceed to generate solution validation until you have confirmed the deployment guidance with the user.
terraform plan or (where supported) gcloud ... --dry-run).curl, ping, or gcloud)
and provide them to the user to execute, to test live endpoint
reachability, networking paths, and load balancer routing.solution-architecture-guide.md,
based on the template in Output Template.In these kits
More from @google
Works with
Claude, Codex, Cursor & more