os-improvement-report

To visualize agentic loop evaluation results, this Skill reads numeric data from experiment logs and ledger sources to generate a timeline chart tracking performance evolution. It produces a visual chart with KEEP/DISCARD markers and a concise summary.

Tracking improvement deltas requires aggregating per-cycle evaluation data from logs and ledger sources. This Skill surfaces keep/discard delta narratives and highlights baselines alongside top improvements in an automated textual summary.

Yes, you can use matplotlib and pandas to chart agentic loop logs. This Skill uses both libraries alongside numpy to read numeric evaluation results, process the data, and produce a visual timeline chart of eval scores.

Yes, the improvement reporting workflow supports exploration-cycle plugins. It is designed for cross-skill applicability, working with agentic-os and related plugins that log results in standard sections to aggregate cycle data.

To generate an automated evaluation report, you need existing per-cycle eval results and ledger data logged in standard sections. The Skill aggregates these sources to output both a visual progress chart and a textual summary.