context-optimization - Optimize context efficiency and capacity
Improves context efficiency through KV-cache optimization, observation masking, compaction, partitioning, retrieval scoping, and token budget management.
Tags
Updated: 2026-09-30Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Optimize KV-cache reuse
- Mask verbose observations
- Compact accumulated context
- Partition work across sub-agents
- Allocate token budgets
- Monitor optimization metrics
Inputs
- Context window limit
- System prompt
- Tool definitions
- Conversation history
- Tool outputs
- Retrieved documents
- Token usage metrics
- Quality baseline
Outputs
- Compacted context
- Masked observation references
- Partitioned subtask results
- Token budget changes
- Optimization measurements
Requirements
- Runtime with prefix caching support
- External storage for retrievable observations
- Sub-agent execution support