This digs through your Agent Monitor data to find patterns you didn't know existed. It pulls tool transition matrices (which tools follow which), error cascades by agent depth, model delegation habits, and compaction triggers across your last 200 sessions. You get a ranked report of the top 10 patterns with frequency counts and whether they're helpful habits or anti-patterns to fix. Most useful after you've accumulated enough session history to make the statistics meaningful. The toolFlow transitions are probably the most immediately actionable part, showing you which tool chains you repeat constantly and where retry loops happen. Think of it as workflow archaeology for your Claude usage.
npx -y skills add hoangsonww/claude-code-agent-monitor --skill pattern-detect --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
sickn33/antigravity-awesome-skills
kubesphere/kubesphere
supercent-io/skills-template