secure-development-policy（安全发展政策）

🛡️ Secure Development Policy Skill

🎯 Purpose Statement

Apply Hack23 AB's Secure Development Policy to demonstrate how security-by-design creates competitive advantages through systematic DevSecOps implementation.

Reference: Hack23 Secure Development Policy

This skill provides comprehensive guidance for implementing all phases of the secure development lifecycle with clear, actionable requirements following STYLE_GUIDE.md icon standards.

🔄 Secure Development Lifecycle (SDLC) - All Phases

📋 Phase 1: Planning & Design

Clear Requirements:

• ✅ Complete project classification per Classification Framework
• ✅ Document CIA triad levels (Confidentiality, Integrity, Availability)
• ✅ Define RTO/RPO aligned with business impact
• ✅ Create SECURITY_ARCHITECTURE.md with Mermaid C4 diagrams
• ✅ Perform STRIDE threat modeling documented in THREAT_MODEL.md
• ✅ Complete risk assessment integrated with Risk Register
• ✅ Calculate security investment ROI based on classification

Deliverables Checklist:

• README.md includes "Project Classification" section
• SECURITY_ARCHITECTURE.md created with current state
• FUTURE_SECURITY_ARCHITECTURE.md created with roadmap
• THREAT_MODEL.md includes STRIDE analysis
• Risk assessment documented and registered
• Cost-benefit analysis approved by CEO

💻 Phase 2: Development

Clear Requirements:

• ✅ Follow OWASP Top 10 secure coding standards
• ✅ Implement language-specific security best practices
• ✅ Require security-focused code review for all changes
• ✅ Apply data classification to all code assets
• ✅ Use GitHub secrets, never hardcode credentials
• ✅ Implement secret rotation procedures
• ✅ Use parameterized queries (no SQL injection)
• ✅ Validate and sanitize all inputs
• ✅ Encode outputs appropriately for context
• ✅ Implement proper error handling (no information disclosure)

Code Review Security Checklist:

• No hardcoded secrets or credentials
• Input validation on all external data
• Output encoding prevents XSS
• Parameterized queries prevent injection
• Authentication/authorization properly implemented
• Error messages don't leak sensitive info
• Logging captures security events
• Dependencies are approved and scanned
• Code follows least privilege principle
• Sensitive data encrypted at rest and in transit

🧪 Phase 3: Security Testing

Clear Requirements:

🔬 SAST (Static Application Security Testing)

• ✅ SonarCloud integration on every commit
• ✅ Quality gate must pass (no Critical/High vulnerabilities)
• ✅ Security hotspots reviewed and resolved
• ✅ Code coverage ≥80% lines, ≥70% branches

📦 SCA (Software Composition Analysis)

• ✅ Automated dependency scanning (Dependabot/Renovate)
• ✅ SBOM generation (CycloneDX or SPDX format)
• ✅ License compliance via FOSSA scanning
• ✅ Vulnerability remediation per classification SLAs:

- Critical: 24 hours - High: 7 days - Medium: 30 days - Low: 90 days

⚡ DAST (Dynamic Application Security Testing)

• ✅ OWASP ZAP scanning in staging environment
• ✅ API security testing (authentication, authorization, input validation)
• ✅ Annual penetration testing for High/Critical projects

🔍 Secret Scanning

• ✅ GitHub secret scanning enabled
• ✅ Pre-commit hooks prevent secret commits
• ✅ Immediate rotation for any exposed secrets

🔒 Test Data Protection

• ✅ NEVER use production data in test environments
• ✅ Anonymize/pseudonymize/mask all test data
• ✅ Securely delete test data after use
• ✅ Restrict test environment access (least privilege)

🚀 Phase 4: Deployment

Clear Requirements:

• ✅ Automated CI/CD with security gates
• ✅ All GitHub Actions pinned to SHA commits
• ✅ Harden-runner enabled with egress auditing
• ✅ Least privilege permissions in workflows
• ✅ Manual approval required for production
• ✅ Deployment checklist completed
• ✅ Security metrics and monitoring configured
• ✅ Rollback plan documented and tested

GitHub Actions Security Pattern:

permissions:
  contents: read  # Least privilege default

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: step-security/harden-runner@<SHA>
        with:
          egress-policy: audit

      - uses: actions/checkout@<SHA>

      # Pin ALL actions to SHA, not tags

🔧 Phase 5: Maintenance & Operations

Clear Requirements:

• ✅ Active vulnerability management per Vulnerability Management
• ✅ Security metrics tracked per Security Metrics
• ✅ Regular dependency updates (automated via Dependabot)
• ✅ Security patches applied per classification SLAs
• ✅ Incident response integration per Incident Response Plan
• ✅ Annual security review and threat model update
• ✅ Quarterly dependency and license audits
• ✅ Continuous monitoring and alerting

Monitoring Requirements:

• Security event logging enabled
• SIEM integration configured
• Vulnerability alerts routed to security team
• Performance metrics track security impact
• Compliance dashboards updated real-time

🤖 AI-Augmented Development Controls

Clear Requirements for ALL AI-assisted development:

🔐 AI as Proposal Generator Only

• ✅ All AI outputs require human review and approval
• ✅ AI cannot bypass CI/CD, security gates, or approvals
• ✅ Human developer retains accountability for all changes
• ✅ AI cannot weaken security controls
• ✅ AI cannot autonomously deploy to production

📋 Mandatory PR Review Process

• ✅ All AI-assisted code goes through standard PR workflow
• ✅ PR description documents AI assistance used
• ✅ Security gates enforced (no exceptions for AI)
• ✅ Human reviewer verifies security controls intact
• ✅ Testing requirements unchanged (80%+ coverage)

🔧 Curator-Agent Governance

• ✅ Changes to .github/agents/*.md require CEO approval
• ✅ Changes to .github/copilot-mcp*.json require CEO approval
• ✅ Changes to .github/workflows/copilot-setup-steps.yml require CEO approval
• ✅ Treat as Normal Changes per Change Management
• ✅ Document risk assessment for capability expansion
• ✅ All new integrations require security review

🛡️ Security & Audit Requirements

• ✅ Agents operate with least-privilege tool access
• ✅ MCP configurations under change control
• ✅ All agent activities logged and auditable
• ✅ Capability expansion requires security review
• ✅ Quarterly agent ecosystem security audit

🎯 Unit Test Coverage Requirements

Clear Requirements:

• ✅ Minimum 80% line coverage
• ✅ Minimum 70% branch coverage
• ✅ Tests run on every commit and PR
• ✅ UnitTestPlan.md documented in repository
• ✅ Public coverage reports via badges
• ✅ Historical coverage tracking (no regression)
• ✅ Coverage reports published to public URL

Reference Implementations:

• 🏛️ CIA: Unit Tests • Coverage • Plan
• 🎮 Black Trigram: Tests • Coverage • Plan
• 📊 CIA Compliance Manager: Tests • Coverage • Plan

🌐 End-to-End Testing Requirements

Clear Requirements:

• ✅ All critical user journeys covered
• ✅ E2ETestPlan.md documented in repository
• ✅ Public Mochawesome/Cypress reports
• ✅ Cross-browser testing (Chrome, Firefox, Safari)
• ✅ Performance assertions within tests
• ✅ Authentication/authorization flows tested
• ✅ Error handling and edge cases covered

Reference Implementations:

• 🏛️ CIA: E2E Plan
• 🎮 Black Trigram: E2E Tests • Plan
• 📊 CIA Compliance Manager: E2E Tests • Plan

🕷️ Threat Modeling Requirements

Clear Requirements per Threat Modeling Policy:

📋 Required Documentation

• ✅ THREAT_MODEL.md in repository root
• ✅ STRIDE framework application (all 6 categories)
• ✅ MITRE ATT&CK technique mapping
• ✅ Attack tree analysis with probability/impact
• ✅ Threat agent classification (external/internal/supply chain)
• ✅ Quantitative risk assessment with business impact
• ✅ Security control mapping with effectiveness validation

🔄 Maintenance Requirements

• ✅ Threat model created during design phase
• ✅ Updated for all architectural changes
• ✅ Annual comprehensive review
• ✅ Quarterly update cycle
• ✅ Incident-driven updates after security events

Reference Implementations:

• 🏛️ CIA: Threat Model • STRIDE • Attack Trees
• 🎮 Black Trigram: Threat Model • Gaming Security
• 📊 CIA Compliance Manager: Threat Model • Risk Assessment

🎖️ Public Evidence Requirements

Clear Requirements per STYLE_GUIDE.md:

Required Badges in README.md

• ✅ OpenSSF Scorecard (target: ≥7.0)
• ✅ CII Best Practices (minimum: Passing)
• ✅ SLSA Level 3 attestation
• ✅ SonarCloud Quality Gate (must be "Passed")
• ✅ Code Coverage badge (≥80%)
• ✅ License badge (OSI-approved)
• ✅ Threat Model link badge

Example Badge Configuration:

[![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/ORG/REPO/badge)](https://scorecard.dev/viewer/?uri=github.com/ORG/REPO)
[![CII Best Practices](https://bestpractices.coreinfrastructure.org/projects/ID/badge)](https://bestpractices.coreinfrastructure.org/projects/ID)
[![Quality Gate](https://sonarcloud.io/api/project_badges/measure?project=ORG_REPO&metric=alert_status)](https://sonarcloud.io/summary/new_code?id=ORG_REPO)
[![Coverage](https://sonarcloud.io/api/project_badges/measure?project=ORG_REPO&metric=coverage)](https://sonarcloud.io/summary/new_code?id=ORG_REPO)
[![Threat Model](https://img.shields.io/badge/Threat_Model-Documentation-blue)](https://github.com/ORG/REPO/blob/main/THREAT_MODEL.md)

📋 Master Verification Checklist

Use this comprehensive checklist before ANY deployment:

Phase 1: Planning & Design ✅

• Project classification complete (CIA + RTO/RPO + Business Impact)
• SECURITY_ARCHITECTURE.md created with C4 diagrams
• FUTURE_SECURITY_ARCHITECTURE.md created with roadmap
• THREAT_MODEL.md includes STRIDE + MITRE ATT&CK
• Attack trees documented with probability/impact
• Risk assessment registered
• Cost-benefit analysis approved

Phase 2: Development ✅

• OWASP Top 10 mitigations implemented
• Language-specific security standards followed
• Code review completed with security focus
• Asset classification applied
• No hardcoded secrets (verified with secret scanning)
• Input validation implemented
• Output encoding implemented
• Error handling doesn't leak sensitive info
• Logging captures security events

Phase 3: Testing ✅

• SAST passed (SonarCloud quality gate green)
• SCA clean (all vulnerabilities remediated per SLA)
• DAST completed (OWASP ZAP scan clean)
• Secret scanning clean
• Test data properly anonymized
• Unit test coverage ≥80% lines, ≥70% branches
• E2E tests cover all critical paths
• UnitTestPlan.md documented
• E2ETestPlan.md documented

Phase 4: Deployment ✅

• CI/CD security gates passed
• All GitHub Actions pinned to SHA
• Harden-runner enabled
• Least privilege permissions
• Manual approval obtained for production
• Deployment checklist complete
• Security metrics configured
• Rollback plan tested

Phase 5: Operations ✅

• Vulnerability management active
• Security metrics monitored
• Dependency updates automated
• Incident response integrated
• Annual review scheduled
• Quarterly audits scheduled

AI Controls ✅

• Human review for all AI outputs
• No autonomous deployments
• PR documentation complete
• Agent permissions least-privilege
• MCP config under change control
• Audit trail enabled

Documentation ✅

• SECURITY_ARCHITECTURE.md
• FUTURE_SECURITY_ARCHITECTURE.md
• THREAT_MODEL.md
• SECURITY.md (vulnerability disclosure)
• WORKFLOWS.md (CI/CD documentation)
• UnitTestPlan.md
• E2ETestPlan.md
• CRA-ASSESSMENT.md (if applicable)
• README includes classification section
• All required badges displayed

Public Evidence ✅

• OpenSSF Scorecard ≥7.0
• CII Best Practices (Passing+)
• SLSA Level 3 attestation
• Quality Gate passed
• Coverage ≥80%
• License compliance (FOSSA clean)

📚 Authoritative References

• Primary Policy: Secure Development Policy
• Related Policies:

- Threat Modeling - Vulnerability Management - Classification Framework - Change Management - Incident Response Plan - Security Metrics

• Standards: STYLE_GUIDE.md

💡 Key Takeaways

• 🛡️ Security by Design: Build security in from the start, not bolt on later
• 🌟 Transparency: Public evidence through badges and documentation
• 🏷️ Classification-Driven: All decisions aligned with project classification
• 🔄 Continuous Improvement: Regular reviews, updates, and learning from incidents
• 🤖 Human Accountability: AI assists, humans decide and approve
• 📊 Evidence-Based: Every security claim backed by public, verifiable proof
• ✅ Comprehensive Coverage: ALL 5 SDLC phases with clear, actionable requirements