Skip to content

Latest commit

 

History

History
75 lines (59 loc) · 4.77 KB

File metadata and controls

75 lines (59 loc) · 4.77 KB

GitHub Copilot Custom Agents

Repository-level Copilot agents for Hack23 / game. Each agent is a YAML-front-matter Markdown file in this directory.

Note: Repository agents do not declare mcp-servers in their YAML — MCP wiring lives in ../copilot-mcp.json and is shared by every agent. All agents here declare tools: ["*"] so they have full access to the tools Copilot exposes; governance is provided by the ISMS AI-Augmented Development Controls (see the ai-augmented-sdlc skill).

Available Agents

Agent Expertise
🎮 game-developer Three.js + @react-three/fiber, game loops, 60 fps, accessibility
🎨 frontend-specialist React 19, strict TypeScript, WCAG 2.2 AA, bundle optimization
🧪 test-engineer Vitest + Cypress + RTL, ≥ 80 % coverage, ≥ 95 % on security code
🔒 security-specialist OSSF Scorecard, SLSA L3, OWASP, STRIDE, ISMS compliance
📝 documentation-writer JSDoc, Mermaid, READMEs, ADRs, C4 models, ISMS citations
🎯 product-task-agent Cross-cutting analysis, issue engineering, Copilot coding-agent orchestration

Agent × Skill Map

Agent Primary skills Secondary skills
🎮 game-developer react-threejs-game, performance-optimization testing-strategy, security-by-design
🎨 frontend-specialist performance-optimization, testing-strategy documentation-standards, security-by-design
🧪 test-engineer testing-strategy react-threejs-game, security-by-design
🔒 security-specialist security-by-design, isms-compliance testing-strategy, ai-augmented-sdlc
📝 documentation-writer documentation-standards, isms-compliance ai-augmented-sdlc
🎯 product-task-agent all 7

ISMS Policy Map

Agent Primary ISMS policies
🎮 game-developer SDP, OSP
🎨 frontend-specialist SDP, Privacy
🧪 test-engineer SDP §Testing
🔒 security-specialist ISP, SDP, OSP, Threat Modeling, Cryptography
📝 documentation-writer ISP §Transparency, SDP §Architecture Documentation
🎯 product-task-agent All policies (cross-cutting)

Workflow

graph TB
  User[Developer request] --> Copilot[GitHub Copilot]
  Copilot --> Select{Select agent}
  Select -->|3D / game loop| GD[🎮 game-developer]
  Select -->|React UI| FS[🎨 frontend-specialist]
  Select -->|Tests / coverage| TE[🧪 test-engineer]
  Select -->|Security / ISMS| SS[🔒 security-specialist]
  Select -->|Docs / diagrams| DW[📝 documentation-writer]
  Select -->|Analysis / plan| PT[🎯 product-task-agent]
  PT -.->|creates issues, assigns| GD & FS & TE & SS & DW
Loading

Usage

@workspace Use the game-developer agent to add a new particle effect
@workspace Ask the test-engineer to raise coverage on useGameState
@workspace Have the security-specialist review the dependency additions
@workspace Let product-task-agent plan the next quality-improvement sprint

Governance

  • Every agent change is a Normal Change under Change Management
  • All agents use tools: ["*"]; least-privilege is enforced by the ai-augmented-sdlc skill and human review
  • MCP server configuration lives in ../copilot-mcp.json and uses secrets.COPILOT_MCP_GITHUB_PERSONAL_ACCESS_TOKEN

Resources