**ROLE ENFORCEMENT IMPLEMENTATION PLAN** I've created 5 subtasks to implement strict role enforcement for Tinker agents: **CRITICAL PATH:** 1. Task 1 (Critical) - System Prompts: Foundation for role constraints 2. Task 2 (High) - Role Validation: Active prevention of violations 3. Task 3 (High) - MCP Access Control: Tool-level restrictions 4. Task 4 (Medium) - Guardrails: Helpful error messages 5. Task 5 (High) - Testing: Comprehensive verification **ROLE BOUNDARIES DEFINED:** **Orchestrator:** Strategic planning, task decomposition, coordination - NEVER implements code directly - Creates and manages tasks only - Uses planning/coordination tools **Worker:** Code implementation, testing, bug fixes - NEVER creates tasks or plans work - Implements assigned tasks only - Uses implementation tools **Reviewer:** Code review, audit, quality assurance - NEVER implements new features - Reviews and validates only - Uses review/audit tools **WORKERS NEEDED:** - 1 Backend developer for Tasks 1-3 (core implementation) - 1 Frontend/UX developer for Task 4 (error messages) - 1 QA engineer for Task 5 (comprehensive testing) All tasks are now ready for workers to claim. The implementation will ensure strict role adherence across all Tinker agents.