codejunkie99/agentic-stack — reverse-engineered prompt

Build me a local first tool that gives me one portable agent brain I can drop into any coding project and keep using even if I switch between tools like Claude Code, Cursor, Codex, Windsurf, Copilot CLI, Gemini, or a simple Python loop.

I want it to create and manage a .agent folder with memory, reusable skills, and shared protocols, then wire that into whichever harness I choose through a simple install or onboarding flow. Please include a friendly first run wizard, commands to add or remove harnesses later, a health check, status view, safe upgrade flow, and a way to repair or resync skill metadata without overwriting my own notes or settings.

Also give me one place to monitor everything locally, including recent agent activity, cron runs, active agents, rough token and cost estimates, KPI style summaries, daily dashboard exports, and local artifact exports from approved redacted runs like traces, eval cases, and JSONL. If it fits, add optional integration with an external long term memory tool for cross project recall. Look up current docs online if needed.