This old command center has been superseded by the new page: ‣
Do not use this page as the active Zeref source. It remains only because the available connector can update pages but cannot permanently delete or trash a page.
This is your guidebook for Zeref OS — a local-first context and memory engine that wraps any AI model and converts your goals into structured, high-quality outputs. It is not a chatbot configuration. It is not a plugin. It is an operating kernel you install once and use across every AI environment you touch.
Current Version: v1.0.0 | Released: June 2026 | GitHub: kanadhiayash/zeref-os | Canonical spec: AGENTS.md
Zeref OS is the harness, not the model. The model is a commodity. The harness — what surrounds the task, structures context, routes work, enforces quality, and compresses sessions — is where lasting value lives.
Three core principles:
1. The harness is the moat. A well-configured harness produces measurably better output than a bare model. Zeref is that harness.
2. Context engineering over prompt engineering. What surrounds a question determines output quality more than how the question is worded. Zeref structures what the model knows before it acts.
3. Self-improvement must be auditable and human-curated. Every upgrade proposal comes from evidence, not self-assessment.
| Your goal | Where to go |
|---|---|
| I am new to Zeref OS | Start Here |
| I want to understand the operating rules | Core Operating Rules |
| I want to set up Claude (any surface) | Claude Main Instructions |
| I want to set up Claude Cowork | Cowork Global Instructions |
| I want to set up coding agents (Codex, Cursor, Windsurf, Copilot, Zed, Junie, Aider, Cline, Roo) | Codex / Coding Agent Instructions |
| I want to set up ChatGPT | ChatGPT Setup |
| I want to set up Gemini or Antigravity | Gemini + Antigravity Setup |
| I want to set up Perplexity | Perplexity Setup |
| I want to set up Replit, v0, Bolt, or Lovable | Prompt-to-App Builders |
| I want the full 39-surface environment index | Cross-Agent Instruction Strategy |
| I want prompts to copy into any environment | Prompt Library |
| I want a step-by-step workflow guide | Workflow Playbooks |
| I want to understand the Caveman handoff protocol | Handoff Protocol |
| I want to track a project or session | Project Tracker / Session Tracker |
| I am troubleshooting or checking safety rules | Troubleshooting & Safety Hub |
| I want to know which model or environment to use | Model Selection Guide |
| I want to configure my positioning | User Positioning |
| Item | Status |
|---|---|
| Version | v1.0.0 |
| Canonical spec | AGENTS.md in kanadhiayash/zeref-os |
| Architecture | Local-first context and memory engine |
| Agents | 6 (Orchestrator, CPO, CTO, CDO, COO, CQO) |
| Core skills | 10 |
| Commands | 8 (start, stop, done, status, team, compress, handoff, fairytale) |
| Harness surfaces covered | 39 paste blocks across 24 environments |
| Memory layout | Flat memory/ (hot.md, index.md, wiki pages) |
| Privacy default | Abstract |
| QA gate | Evidence Discipline + pre-output checklist |
| Last revamp | June 2026 |