What problem does it solve?
AI assistants have limited context windows and every token carries cost and attentional overhead; this Skill helps teams plan and enforce token budgets, prevent context rot, and keep model outputs accurate and cost-effective across sessions and multi-file workflows.
Core Features & Use Cases
- Token Budgeting & Accounting: Measure token consumers, estimate costs, and produce a practical token budget for a task or project.
- Context Window Management: Apply progressive disclosure, lightweight identifiers, and distraction filtering to surface high-signal tokens only when needed.
- Caching & Compaction Guidance: Identify cacheable prefixes, compute cache break-even points, and recommend automatic compaction thresholds.
- Agent Orchestration Patterns: Design subagent isolation, parallel execution, specialization by role, and lifecycle hooks (PreToolUse, PostToolUse) for deterministic checks.
- Use Cases: Planning multi-file code changes with a tight token budget, optimizing prompts for iterative review, diagnosing degraded outputs in long sessions, and designing orchestration for multi-agent systems.
Quick Start
Use the context-engineering skill to analyze your project's open files, produce a token budget, and recommend which files to pin, trim, or summarize to meet a specified context target.
