Systems & Frameworks
Where data models, design systems, and AI tooling meet.
I build frameworks that connect canonical data models, semantic tokens, and agent workflows—so teams can move from research to production with less drift and more trust.
Mission Protocol + CMOS
Creator In Progress Mission, memory, and orchestration framework
YAML-first mission and memory model that coordinates human and agent work across research, build, and governance loops.
How it helps teams
- Mission specs that turn ambiguous work into explicit goals, constraints, and success criteria.
- CMOS (Context Management Operating System) for project memory, session logging, and traceability.
- Evaluation hooks for multi-agent systems: task completion, tool selection, context preservation, collaboration quality.
- Patterns for running research → plan → build loops with auditable artifacts instead of opaque prompts.
Agent orchestrationEvaluationGovernance
OODS Foundry
Creator Active Bridging the semantic gap between AI agents and enterprise design systems
A trait-based design system architecture where business objects—not visual components—are the organizing principle. OODS closes the semantic gap that prevents AI agents from truly understanding design systems.
How it helps teams
- Trait-based composition: behaviors like Stateful, Timestampable, and Billable mix into any business object
- Four-layer token stack: primitive → semantic → component → object tokens with full traceability
- Object-oriented organization around User, Subscription, Invoice—not Button, Card, Modal
- Machine-readable schemas that AI agents can parse, validate, and compose
Design systemsObject modelsSemantic tokensAI agentsTrait composition
OODS Visualization
Creator Active Trait-driven data visualization for design systems
A visualization system that extends OODS trait composition to charts, maps, and network diagrams. Dual-renderer architecture (Vega-Lite + ECharts) with a portable Normalized Viz Spec and WCAG 2.1 AA accessibility built in.
How it helps teams
- 18 visualization traits covering marks, encodings, layouts, and interactions
- Dual-renderer architecture supporting Vega-Lite and Apache ECharts
- Normalized Viz Spec as a portable, declarative schema for any chart
- Spatial module for choropleth maps, markers, and geo-data resolution
- Network/flow diagrams: Sankey, treemap, sunburst, force graphs
Data visualizationDesign systemsTrait compositionAccessibilityChartsMaps
Autonomous knowledge system for research and personal knowledge
A research repository that turns fragmented research assets into structured, reusable knowledge. Mission-driven research protocol, evidence-level traceability, and quality-aware search through PEDR's 6-layer hybrid architecture.
How it helps teams
- Mission Protocol for structured research with explicit quality gates
- PEDR: 6-layer hybrid search fusing lexical, semantic, syntactic, pragmatic, governance, and graph layers
- DeepSearch integration for autonomous research agent pipelines
- Evidence traceability from insights back to source chunks with citations
ResearchKnowledge managementSemantic searchUX researchRAGHybrid search
MCP aggregation with semantic routing
An MCP aggregator that combines multiple MCP servers into a single interface with semantic tool routing. Import configs from Claude Desktop, VS Code, or Cursor and serve via stdio or SSE.
How it helps teams
- Aggregate multiple MCP servers behind one interface
- Import existing configs from Claude Desktop, VS Code, Cursor
- Semantic routing to direct tools to the right backend
- SSE server option for web client integration
MCPAI agentsTool aggregationClaudeVS Code
Context + Mission Orchestration for AI-assisted delivery
A project management system optimized for AI agents. Models work as missions, organizes sprints, and maintains project memory through contexts and strategic decisions—all exposed via MCP for seamless agent integration.
How it helps teams
- Mission lifecycle: Queued → Current → In Progress → Completed with dependency tracking
- Context management: project state + master context for long-term memory
- CMOS-MCP server exposes SQLite-backed tools to AI agents
- Orchestration patterns: rsip, delegation, boomerang checkpoints
Project managementAI agentsMCPMissionsContext