Per-agent charts
Claude Code and OpenAI Codex are tracked and rendered separately, so styles never blur together.
Scan historical assistant messages from Claude Code and OpenAI Codex, classify the leading kaomoji with BrightWrapper, and see how each agent's expressions change over time.
Claude Code and OpenAI Codex are tracked and rendered separately, so styles never blur together.
Switch between count and proportional views to compare shifts by day.
Reuse cached classifications and only send unseen messages to BrightWrapper.
Scan historical Claude Code and OpenAI Codex transcripts, classify assistant message prefixes with BrightWrapper, and chart the exact kaomoji that were found — separately per agent.
Run scan walks backward through the newest messages of each agent, up to the limit per agent. Clicking a missing timeline gap scans only that agent and date range, up to the same limit. Charts always redraw from the classification cache.