context-curator

To compress a large codebase context, this skill extracts essential signatures and modules to generate minimal, task-focused context packages for planning, debugging, and handoffs. It prevents token waste by retaining only necessary artifacts.

Context curation for reasoning agents is the process of prioritizing and compressing codebase context into Tier 1-3 curated artifact sets. This ensures deterministic decision-making by providing LLMs with clear, minimal scope.

Yes, you can create task-focused context packages for pipeline handoffs. The skill compresses codebase context into minimal artifacts with clear scope, specifically targeting planning, debugging, and handoff tasks.

Yes, context compression works for large codebases to prevent token waste. It targets large repositories and compresses context into essential signatures and modules, producing minimal packages that prevent context bloat for reasoning agents.

The best way to prioritize code context for LLM routing is to compress the codebase into a Tier 1-3 curated artifact set. This retains only necessary artifacts, ensuring deterministic decision-making with clear scope for reasoning agents.

You need curated artifacts for codebase analysis when reasoning agents operate on large repositories and risk context bloat. If you are performing planning, debugging, or handoffs, curating artifacts ensures deterministic decision-making with clear scope.