List view
## The Vision Transform HexFrame from a knowledge mapping tool into an AI orchestration platform where the hexagonal formalism directly drives multi-agent workflows, context management, and system creation. ## Core Innovation: HexFrame as AI Operating System ### The Jay Agent Architecture **Jay** = The system creation agent that helps users build and improve systems using HexFrame formalism: ``` Center: Help user create a system ├─ 1. Clarify: What system? → [know/ask user] ├─ 2. Foundation: Existing base? → [create/build upon] ├─ 3. Activation: Already using? → [find MVP/continue] ├─ 4. Validation: Serving goal? → [adjust/proceed] ├─ 5. Timing: Improve now? → [yes/reflect] └─ 6. Evolution: How to improve? → [execute] ``` Each step has decision gates: does Jay know the answer or ask the user? ### HexFrame Formalism for AI Orchestration **1. Plans as Frames** - **Goal** = Center tile (what we want to achieve/decide) - **Steps** = Child tiles (how to achieve the goal) - **Context** = Parent/sibling tiles (relevant background) **2. Color-Based Context Engineering** - **Same color siblings**: Single agent maintains context across tasks - **Tint inheritance**: Parent reasoning about child context inclusion - **Lightness gradients**: Context filtering/abstraction levels **3. Edge-Based Interaction Specs** - **Sibling edges**: Data flow, synchronization points - **Parent-child edges**: Context inheritance rules, abstraction boundaries **4. Tile Composition for Reusable Context** - Systems can reference shared context tiles - Version-controlled system knowledge - Clear dependency graphs for complex orchestration ## Implementation Phases ### Phase 1: Prompts as Executable Tiles (Immediate Value) **Goal**: Make tiles executable with "Run" buttons - Tiles containing prompts become runnable - Children provide context for execution - Immediate value: persistent, reusable, shareable prompts - **Differentiation**: Unlike ChatGPT/Claude, prompts are persistent, composable, version-controlled ### Phase 2: Jay Integration & Tool Use **Goal**: Implement Jay's system creation workflow - Add CRUD tools for tiles from chat - Jay creates/updates tiles as it helps users - Emerging maps ARE the system documentation - **Value**: AI that builds structured documentation while helping ### Phase 3: Multi-Agent Router **Goal**: Route conversations to specialized agents ``` Center: Hexframe Assistant ├─ Jay: System creation and improvement ├─ Explorer: Navigate existing systems ├─ Teacher: Explain HexFrame concepts └─ Builder: Technical implementation help ``` ### Phase 4: Advanced Orchestration **Goal**: Full HexFrame formalism leverage - Color-based agent routing and context management - Edge-based workflow orchestration - Tile composition for complex system interactions - Multi-iteration agent workflows with persistent plans ## Success Criteria **Immediate (Phase 1-2)** - Users can execute prompts directly from tiles - Jay helps users build systems while creating documentation - Value proposition clear vs existing AI tools **Medium-term (Phase 3-4)** - Multi-agent conversations feel natural and productive - Users create complex systems through conversation - HexFrame becomes the interface layer for AI orchestration **Long-term Vision** - System thinkers become visionaries through AI amplification - HexFrame formalism proves as the optimal AI coordination protocol - Users build living systems, not just documentation ## The Transformation From: "Knowledge mapping tool with chat" To: "AI orchestration platform where spatial relationships drive agent behavior" The hexagonal map stops being documentation and becomes the computational substrate for AI coordination - the visual representation IS the execution plan. ## Dependencies - **Milestone 1**: Must complete dogfooding to validate workflow approach - **MCP Integration**: Required for tile CRUD from chat - **User Experience**: Simple onboarding for non-technical users This milestone represents the core innovation that makes HexFrame fundamentally different from existing AI tools - turning spatial organization into computational orchestration.
No due date## The Goal Complete the transition from file-based workflow to HexFrame tiles, proving that systems can stay alive through AI-readable structure and actual daily use. ## Why This Matters ### The System Thinker's Dream We've built a workflow system that should guide development. But like all systems created by system thinkers, it risks dying unused. This milestone tests whether we can break that pattern by making the workflow system native to HexFrame itself. ### The Innovation: Spatial Workflow Management Instead of scattered files, workflow lives in HexFrame tiles that AI can naturally read and update: - Workflow structure visible as hexagonal map - Current state maintained in tiles, not files - AI gets milestone context when working on priorities - Progress tracked through tile updates, not file changes ## Success Criteria 1. **Complete Migration**: All `.workflow/` files replaced by HexFrame tiles 2. **MCP Integration**: Claude can read/update workflow through MCP tools 3. **Executable Prompts**: Key workflows accessible via `@tile` syntax and MCP 4. **Living System**: Workflow state stays current through natural usage 5. **Transparent Process**: Development progress visible and trackable ## HexFrame Value Demonstration This milestone proves HexFrame's core value propositions: ### 1. Executable Prompt Library - **@deploy**: Production deployment automation (merge develop→main, create issues, archive cycle) - **@context**: Get structured context for current priority (milestone + cycle + priority details) - **@health**: System aliveness metrics (MCP usage, prompt execution frequency) - **@expert**: HexFrame system creation expertise (how to make systems that live) ### 2. Structured Data Access - **HexFrame UI**: Visual workflow navigation for humans - **MCP Integration**: Programmatic access for AI agents - **Progressive Context**: Right information at right abstraction level ### 3. System Composability - Named tiles become reusable components - Systems can reference other systems - Import/customize workflow patterns across users ### 4. Aliveness Metrics - Track which prompts get executed - Monitor MCP API usage patterns - Measure system engagement vs abandonment ## Concrete Workflow Structure Simplified hierarchy optimized for solo development with AI context: ``` Goals (Center) ├─ 1. Current Milestone │ ├─ 1.1. Current Cycle │ │ ├─ 1.1.1. Priority 1 │ │ ├─ 1.1.2. Priority 2 │ │ ├─ 1.1.3. Priority 3 │ │ ├─ 1.1.4. Priority 4 │ │ └─ 1.1.5. Bonus │ ├─ 1.2. Next Cycle │ ├─ 1.3. Backlog (milestone) │ └─ 1.4. Archive ├─ 2. Next Milestone ├─ 3. Backlog (global) └─ 4. Milestones Archive ``` **Context Hierarchy**: Goals > Milestone > Cycle > Priority - Each level provides necessary context for the next - AI agents get progressive context building - Human navigation follows natural decision paths ## Implementation Features ### @ Syntax for Executable Prompts - **Chat Interface**: `@deploy` executes deployment workflow - **MCP Tool**: `hexframe:ask(coordinates, question)` for programmatic access - **Named Systems**: Compose multiple tiles into named, executable systems ### Automation Ready - **Cycle Completion**: Automated GitHub issue creation, merge, documentation - **State Transitions**: Move completed cycles to archive automatically - **Metrics Collection**: Track system usage and prompt execution ## Validation Strategy ### Week 1: Structure & Migration - Create Goals tile with 4 children (milestone, next milestone, backlog, archive) - Populate Current Milestone > Current Cycle > Priorities - Implement `@` syntax for key prompts (@deploy, @context, @health) - Update CLAUDE.md to use MCP instead of file reading ### Week 2: Daily Workflow Usage - Use tile-based workflow for actual development - Execute priorities using AI guidance from tile context - Track prompt usage and system engagement - Validate automation works (cycle completion, archiving) ### Week 3: Refinement & Metrics - Optimize tile organization based on usage patterns - Implement aliveness metrics and reporting - Document lessons for next milestone - Prepare transition to Jay system architecture ## Success Metrics - **Zero `.workflow/` files**: All state lives in tiles - **AI Effectiveness**: Claude accurately guides development using tile context - **Prompt Usage**: @deploy, @context executed regularly in actual development - **System Evolution**: Workflow structure improves through practice ## The Meta-Demonstration This milestone becomes HexFrame's first success story: - **Living System**: Workflow that stays current through use, not neglect - **AI Integration**: Spatial organization drives AI understanding - **Executable Documentation**: Systems that work, not just describe - **Recursive Development**: Using HexFrame to build HexFrame Success here validates the foundation for Milestone 2's Jay system and full AI orchestration platform.
No due date•1/2 issues closed