AgentSkillsCN

project-standards

按需提供项目编码标准与架构原则。在实现功能、编写代码或做出技术决策时自动调用。读取constitution.md以确保所有代码均符合项目原则。

SKILL.md
--- frontmatter
name: project-standards
description: Provides project coding standards and architectural principles on-demand. Automatically invoked when implementing features, writing code, or making technical decisions. Reads constitution.md to ensure all code aligns with project principles.
allowed-tools:
  - Read
  - Grep
  - Glob

Project Standards Skill

You are implementing code in a project that follows specific architectural principles and coding standards defined in the project constitution.

When to Invoke This Skill

Claude should use this skill when:

  • •Writing or modifying code
  • •Making architectural decisions
  • •Implementing tasks from tasks.md
  • •Reviewing code for compliance
  • •Setting up new components or services

How to Use

  1. •

    Read the constitution to understand project principles:

    code
    Read: .specify/memory/constitution.md
  2. •

    Apply principles during implementation:

    • •TDD: Write tests first, ensure they fail, then implement
    • •SOLID: Follow single responsibility, dependency injection
    • •Security: No hardcoded secrets, validate all inputs
    • •Observability: Add logging, metrics, tracing
  3. •

    Check compliance before marking tasks complete

Key Principles Summary

Test-Driven Development (TDD)

  • •Write tests BEFORE implementation code
  • •Tests must FAIL initially (Red phase)
  • •Implement to make tests pass (Green phase)
  • •Refactor after passing (Refactor phase)
  • •100% test pass rate required (enforced by test-gate.sh)

SOLID Architecture

  • •Single Responsibility: One reason to change per class/module
  • •Open/Closed: Extend via interfaces, don't modify
  • •Liskov Substitution: Implementations substitutable for abstractions
  • •Interface Segregation: Many focused interfaces
  • •Dependency Inversion: Depend on abstractions, inject implementations

Security-First

  • •Secrets in environment variables or secret managers only
  • •Input validation with strict schemas (Pydantic, Zod)
  • •Parameterized queries only (no string interpolation for SQL)
  • •Row-Level Security (RLS) for multi-tenant data
  • •Rate limiting on all API endpoints

Code Quality

  • •Type hints/annotations required
  • •Linting must pass (ruff, ESLint)
  • •Docstrings for public APIs
  • •Meaningful variable/function names

Observability

  • •Structured JSON logging with correlation IDs
  • •Health check endpoints (/health, /ready, /live)
  • •Metrics for critical operations
  • •Distributed tracing for cross-service calls

Integration with Sub-Agents

When working as a specialized sub-agent:

For backend-api-engineer

  • •Read constitution for API versioning rules
  • •Apply security requirements to endpoints
  • •Follow TDD for all endpoint implementations

For frontend-react-engineer

  • •Read constitution for accessibility requirements
  • •Apply performance budgets
  • •Follow component architecture patterns

For openapi-spec-author

  • •Read constitution for API stability rules
  • •Ensure backward compatibility
  • •Follow versioning strategy

For integration-tester

  • •Read constitution for test coverage requirements
  • •Verify compliance with success criteria
  • •Check observability is in place

Quick Reference

Before implementing, ask yourself:

  1. •Did I write the test first? (TDD)
  2. •Is this class doing only one thing? (SRP)
  3. •Are secrets externalized? (Security)
  4. •Is there logging/metrics? (Observability)
  5. •Are inputs validated? (Security)
  6. •Does this align with the constitution? (Compliance)

Full Constitution Access

For complete details, read:

code
.specify/memory/constitution.md

This contains:

  • •All 8 core principles with detailed requirements
  • •Security requirements and data protection rules
  • •Compliance standards (SOC 2, HIPAA, GDPR)
  • •Performance targets and SLOs
  • •Development workflow and CI/CD gates