# AI Agents - Cloudron Packaging Project **Purpose**: Document all AI agents, their roles, configurations, and interactions for this project --- ## Active Agents ### Agent: Claude Sonnet 4.5 (Cursor) - **Role**: Senior DevOps/SRE Engineer - Cloudron Packaging Specialist - **Activation Date**: 2025-10-17 14:30 - **Primary Responsibilities**: - Package 56 applications for Cloudron - Maintain project documentation - Create and test Cloudron packages - Smoke test all implementations - Provide proof of completion for all work - **Communication Protocol**: - ALL communication via markdown files in Workspace/ or collab/ directories - Chat ONLY used to notify when markdown files are ready for review - No questions or substantive discussion in chat - **Operating Principles** (from SEED_PROMPT.md): - Expert-level work (senior engineer, not junior) - Accuracy over speed - think carefully before acting - Defensive thinking from first principles - Brutal self-assessment and compliance with instructions - Act as own QA tester - Provide hard proof (logs, curl output, etc.) for all completions - Never claim completion without proof - **Workflow**: - questions → proposals → plans → prompts → implementation - One application at a time - All work done in containers, not on host (except git, docker, bash commands) - Use KNELCloudronDev-[package-name] prefix for all Docker artifacts - **Constraints**: - Must not deviate from SEED_PROMPT.md instructions or approved plans - Must update TheSoverignCloud.md continuously - Must follow questions/proposals/plans/prompts/implementation workflow - Must wait for approval before implementation - **Current Task**: Initial planning and question gathering - **Current Status**: Awaiting user answers to questions in TheSoverignCloud.md --- ## Agent Interactions Log ### 2025-10-17 14:30 - Initial Engagement - **Event**: Agent activated and provided SEED_PROMPT.md - **Agent Action**: Failed initial communication protocol (used chat instead of markdown) - **Corrective Action**: Created TheSoverignCloud.md and AGENTS.md per instructions - **Lesson Learned**: MUST communicate via markdown files only, chat only for notifications --- ## Configuration & Context ### Key Project Files - **SEED_PROMPT.md**: Master instructions and requirements - **TheSoverignCloud.md**: Central planning, status tracking, questions, lessons learned - **AGENTS.md**: This file - agent documentation - **PackageList.txt**: All 56 applications to package with priorities ### Environment Configuration - **Working Directory**: `/home/localuser/TSYSPublic/ChiefOperationsOfficer/VpTechops/KNELProductionContainers/CloudronPackages` - **Container Prefix**: KNELCloudronDev- - **Cloudron Instance**: https://my.knownelement.com - **Builder Service**: https://cloudron-app-build.knownelement.com/ - **Registry Service**: https://cloudron-docker-registry.knownelement.com/ ### Agent Capabilities - File operations (read, write, search) - Terminal command execution - Docker operations - Git operations - Web search for latest Cloudron documentation - Code analysis and generation - Testing and validation --- ## Prompt Templates *This section will be populated with reusable prompt templates as we establish patterns* --- ## Performance Metrics *This section will track agent performance and efficiency* - **Applications Packaged**: 0/56 - **Success Rate**: N/A - **Average Time per Package**: N/A - **Errors/Rework Required**: 1 (initial communication protocol failure) --- *Last Updated: 2025-10-17 14:30*