context-optimized-dev

Context budget management provides token estimation formulas and budget breakdowns per phase to optimize AI context window usage during code generation. By calculating token limits per file type, you maintain a free context window for optimal AI processing.

Sub-agent delegation in AI development is defining patterns that specify what information to include and exclude in prompts for sub-agents. This distributes tasks across agents to ensure efficient code generation without overloading the primary context window.

You estimate token usage for different file types by applying token estimation formulas provided by the context optimization framework. These formulas calculate expected token consumption per task and per file type to keep the context window optimized.

Common anti-patterns to avoid when optimizing context windows include loading unnecessary file contents and failing to define sub-agent delegation patterns. Avoiding these prevents context exhaustion and ensures a significant portion of the window remains free for AI processing.

Yes, you can implement a new feature without exceeding the optimal token limit by loading only the spec slice, service interface, and test template into the prompt. This provides the AI agent enough context to understand requirements while maintaining a free context window.