AgentSkillsCN

brainstorm-feature

针对全新模块或功能的深度调研与架构设计工作流。严禁提供任何代码示例。适用于在正式实施前,进行高层次的技术设计与研究工作。

SKILL.md
--- frontmatter
name: brainstorm-feature
description: Workflow for deeply researching and architecting a new module/feature. STRICTLY NO CODE EXAMPLES. Use this for high-level technical design and research before implementation.

Brainstorm & Research Feature Workflow (Deep Dive)

This skill guides you to produce a Production-Ready Technical Design for ANY technology stack (Web, Mobile, Backend, IoT).

CRITICAL RULE: Do NOT include code snippets. Use Architecture Diagrams (Mermaid), Schemas, and Abstract Algorithms.

Workflow Steps

1. Requirement Analysis (The "What")

  • Context Detection: Identify the project type (e.g., SaaS, E-commerce, Internal Tool) and stack (SQL/NoSQL, Monolith/Microservices).
  • Actors & Roles: Define exact actors (e.g., Tenant Admin, System, End User, API Client).
  • Invariants: Define hard constraints (e.g., "Non-negative integers", "Immutable audit logs", "Idempotency keys required").

2. Global Benchmarking & Intelligence (Web Research)

  • Search Strategy: Use google_web_search with high-intent enterprise keywords (e.g., "High Availability", "Multi-tenancy isolation", "Audit Trail standards", "Data Sovereignty", "Disaster Recovery patterns").
  • Exclusion Filter: STRICTLY AVOID entry-level tutorials, "How-to" blog posts for beginners, and hobbyist projects.
  • Target Sources:
    • Engineering Blogs: Netflix, Uber, Airbnb, Stripe, Meta, or Stack specific (e.g., Laravel News, Flutter Engineering).
    • Architecture Centers: AWS Architecture Center, Azure Well-Architected Framework, Google Cloud Architecture.
    • Security & Compliance: SOC2/ISO 27001/GDPR/HIPAA requirements relevant to the feature.
  • Fetch: Use read_url_content to extract deep architectural decisions, trade-offs, and scaling bottlenecks.
  • Synthesis: Focus on robustness, security-first design, and operational excellence.

3. Architecting the Solution (The "Blueprint")

Produce a Detailed Design containing:

A. Data Model / Schema

  • Structure:
    • Relational: Table Name, Columns, SQL Types, Indexes, FK Constraints.
    • NoSQL/Document: Collection Name, Document Structure, Sharding Keys.
  • Relationships: Cardinality (1:1, 1:N, M:N) and cascading rules.

B. State Management & Logic

  • State Machine: Transitions and guards described logically (use Mermaid State Diagram if complex).
  • Business Rules: Plain English description of validation and logic (e.g., "If X is active, Y cannot be deleted").

C. Interface / API Surface

  • Definition:
    • REST: Method (GET/POST), Path, Request/Response Body keys.
    • GraphQL: Query/Mutation names, Inputs, Types.
    • Internal: Service Method Signatures (Abstract).
  • Validation Rules: Abstract rules (e.g., "required", "email format", "max length").

D. Performance & Scalability

  • Measured Outcomes: Target latency, throughput, or caching hit ratios.
  • Concurrency: Locking strategies (Optimistic vs Pessimistic), Job Queues, Event sourcing.

4. Handoff

  • Output: Technical Design Brief.
  • Compliance Check: Ensure ZERO code blocks are present.