Ready-to-use templates for implementing The Unbroken Method in your AI-assisted development workflow.
Best for: Getting started immediately, trying out the methodology, simple projects
A minimal, one-page prompt covering the essential principles. Copy, paste, and start working.
Best for: Serious projects, teams adopting the methodology, complex multi-session work
Complete prompt with all seven pillars, anti-patterns, session protocols, and customization guidance.
Best for: Multi-session projects, team handoffs, context preservation
Templates for the four-document handoff protocol: CONTINUATION_PROMPT, AGENT_PLAN, FEATURES, and CHANGELOG.
Best for: Building your failure catalog, team learning, preventing repeated mistakes
Structure and examples for documenting anti-patterns as you discover them.
Best for: Programmatic checkpoints, terminal-based workflows, automation
Bash and Python scripts for implementing collaboration checkpoints in your workflow.
| Pillar | Principle | Key Practice |
|---|---|---|
| 1. Continuous Context | Never break the conversation | Checkpoints + structured handoffs |
| 2. Complete Ownership | If you find it, you fix it | No "out of scope" or "separate issue" |
| 3. Investigation First | Understand before acting | Read, search, test, then implement |
| 4. Root Cause Focus | Solve problems, not symptoms | Ask "why?" until you find the cause |
| 5. Complete Deliverables | Finish what you start | No partial solutions or TODOs |
| 6. Structured Handoffs | Perfect context transfer | Four-document protocol |
| 7. Learning from Failure | Codify mistakes | Anti-pattern catalog |
The Unbroken Method uses two files: an instructions file (methodology) and a project reference (project scope). The instructions file location varies by tool; the project reference is typically AGENTS.md in your repository root.
CLIO loads both files automatically:
.clio/instructions.md - Methodology (The Unbroken Method)AGENTS.md - Project reference (architecture, conventions, commands)CLIO also implements checkpoint discipline, long-term memory, todo tracking, and session persistence natively.
Save your methodology prompt as .github/copilot-instructions.md. Create AGENTS.md in your repository root for project context.
Add methodology to your .cursorrules file. Create AGENTS.md in your repository root for project context.
Add methodology as project instructions in the Claude web interface. Upload AGENTS.md as project knowledge.
Use methodology as custom instructions or system prompt. Paste AGENTS.md content at session start.
Most AI assistants support custom instructions or system prompts. Adapt the templates to your tool's format. The pattern is always the same: methodology instructions tell the AI how to work, and AGENTS.md tells it what the project looks like.