Projects

Production workflow systems growing from professional film and post-production ground truth

• ·Turns non-standard service work into reusable SOPs and productized workflows
• ·Uses professional post-production ground truth as the constraint source
• ·Bridges film, advertising, and post-production judgment with AI-native operations
• ·Designed for repeatable delivery, auditability, and production-grade automation

AI-native development environment with memory, PTY management, and MCP tools

• ·Three-pillar architecture: memory, control, tools
• ·Event-driven session lifecycle management
• ·PTY state as UI authority, JSONL as data authority
• ·Claude Code callback tools via MCP protocol

Operating-system layer for AI coding agents with semantic terminal parsing, persistent memory, and 71 MCP tools

• ·Slot-based management: 1 foreground + N background coding-agent processes
• ·Semantic terminal: Alacritty emulation + regex fingerprint state detection
• ·Knowledge flywheel: hybrid search, typed memories, and retrospective learning loops
• ·71 MCP tools across 4 domains: Knowledge, Compute, Communication, System