context-load

To optimize subagent prompts and reduce token bloat, extract role-specific documentation slices from project memory files like CLAUDE.md and AGENTS.md. This dynamically trims irrelevant information, ensuring the subagent receives only the focused context needed for task performance.

The best way to pass relevant documentation to a specific agent role is by extracting targeted snippets based on role-specific keywords. This approach automatically trims and truncates content to fit within specified token limits while preserving source integrity.

Token budget management for subagent prompt assembly works by automatically trimming and truncating extracted documentation slices to fit within specified token limits. This preserves source integrity while ensuring the assembled prompt remains within the defined maximum token budget.

Yes, you can load context for a database engineer subagent without exceeding a token limit by extracting only the schema and migration documentation from project memory. The system automatically trims the content to fit within your specified maximum token limit.

You should use role-specific context extraction instead of passing full documentation when orchestrating complex tasks where token efficiency and context relevance are critical. It prevents subagents from processing irrelevant information, keeping them focused and efficient.

Validation against defined agent roles is required when extracting documentation slices for multi-agent workflows. The extraction process must also adhere to untrusted content wrapping standards to ensure safe and accurate prompt assembly.