Summary
Highlights
Monitoring and Setup00:01:36
Learn how to use /usage, /context, and status line customization to track token consumption accurately. Understand how memory files, system prompts, and MCP servers affect your initial token usage.
Core Optimization Techniques00:04:32
Implement best practices such as using /clear for new tasks, setting model/effort levels at the start of a chat, and using /rewind instead of /compact to avoid redundant token consumption.
Efficiency Hacks00:10:02
Utilize handoff files for context management between sessions, use ASCII diagrams for UI design to avoid expensive code generation iterations, and keep the .cloudd file concise like an index.
Advanced File and Hook Handling00:17:23
Reduce token usage by using hooks to extract text from PDFs instead of reading the entire document and applying directory-specific .cloudd files to keep the main prompt efficient.
Tool Selection and Automation00:21:27
Prioritize CLI over MCPs when possible, use code graphs for large repositories (500+ files), and prefer deterministic scripts over AI generation for repetitive tasks.