context-health

Context window bloat is identified by inventorying conversation blocks and classifying them as Signal, Bloat, or Overhead. This analysis computes a Bloat Index and staleness distribution to highlight top bloat contributors, generating a Context Health Report with optimization recommendations.

Conversation context staleness measures how outdated a conversation block is relative to the current task. Assessing block staleness alongside token distribution identifies which blocks remain actively useful, isolating bloat that wastes tokens and degrades context window efficiency.

Inventory conversation blocks by labeling each with its type, turn, and a concise description. The analysis then computes task relevance and token distribution across Signal, Bloat, and Overhead categories to generate a detailed Context Health Report.

Analyze context health before long sessions or prior to critical tasks. Running a context health check at these points ensures the conversation context remains clean and efficient by isolating top bloat contributors and generating actionable reduction recommendations.

The best way to identify top bloat contributors is to compute a Bloat Index across all visible conversation blocks. This assesses block staleness and task relevance, categorizing blocks as Signal, Bloat, or Overhead to isolate contributors and generate reduction recommendations.

Yes, context health analysis works without external dependencies. It directly inventories visible conversation blocks, computes a Bloat Index, and assesses staleness and task relevance natively to produce a comprehensive Context Health Report.