Analyzes context window usage trends and audits configuration files for token bloat.
npx skills add https://github.com/greglas75/zuvo --skill context-audit-greglas75
Please help me install this Agent Skill. Skill: context-audit Source: https://github.com/greglas75/zuvo/tree/main/skills/context-audit Command: npx skills add https://github.com/greglas75/zuvo --skill context-audit-greglas75
High-intent search queries and answers about installing and using this skill.
Run the context audit in trend mode to parse runs.log include data collected by the track-includes hook. It shows per-include loading frequency, cumulative byte costs, and the heaviest skill runs so you can identify the largest consumers.
Use the full audit mode, which requires /context output, to evaluate every memory file against five filters: default behavior, contradiction, redundancy, bandaid fixes, and vagueness. It flags files exceeding 200 combined lines and deducts score points per issue.
No, include sizes are captured automatically by the track-includes.sh PostToolUse hook and stored in runs.log field 12 as name:bytes pairs. Only the optional full audit mode requires you to run /context manually first.
If fewer than three runs.log entries contain include data, the audit stops and asks you to run a few skills first so the hook can collect measurements. Trend deltas additionally require more than 14 days of data.
Yes, in full mode it offers to apply settings.json and permissions.deny fixes automatically since they are safe and reversible. Changes to CLAUDE.md are shown as diffs first and require your confirmation before modification.
shekerkammaIdentify missing analytics data points and generate instrumentation requests.
schmitthubAudits agent instruction files, memory stores, and project docs for staleness, symbol errors, and context budget overruns.
tmalcolm-0607Capture quantitative learning signals from AI council channel retrospectives.
EndUser123Parse Claude Code compaction logs to quantify token usage and context size trends.
