agent-density-auditor

To reduce context rot in Claude Code agent prompts, audit agent prompt density to enforce line limits and identify oversized sections. This process measures body lines and estimated tokens to keep agents lean, focused, and reliable.

Agent prompt density measures the volume of text, body lines, and estimated tokens within Claude Code agent files. High density causes context rot, hallucinations, and drift, making prompt density measurement critical for efficient token usage and agent reliability.

To generate a migration plan for an oversized Claude Code agent, analyze the agent prompt to classify sections and recommend content to keep, extract into references, or transform into reusable skills, ensuring compliance with density thresholds.

Yes, you can use Python standard-library tooling and shell scripts to run validation gates after agent edits. This enforces line limits, description length, and frontmatter conventions, returning compliance exit codes for VibeFlow project maintenance.

Enforcing VibeFlow ADR-029 density thresholds is limited to auditing agent bodies, skill files, bootstrap prompts, and descriptions. It requires shell and Python standard-library tooling, and focuses strictly on line limits and section-size warnings rather than semantic logic.