Review fired CCAM alerts and manage rules that watch token thresholds, event patterns, inactivity, or status duration. Create, edit, or delete rules, check cooldown settings, acknowledge alerts, and connect rules to webhook targets.

中文简介正在准备中。当前显示英语简介。

• Reviewing alerts that have not been acknowledged
• Creating a token threshold or event rule
• Changing a rule or checking its cooldown

Review Agent Monitor session data for unusual costs, durations, error rates, token use, event patterns, and behavior. The skill compares sessions with statistical baselines, classifies findings by severity, and reports observed values, possible explanations, and suggested actions.

中文简介正在准备中。当前显示英语简介。

• Spot sessions with unusually high costs or durations
• Find error clusters or repeated tool failures
• Review token and behavior patterns across sessions

Investigate APIError events in Agent Monitor data. The skill tracks error counts over time, identifies affected sessions and models, and classifies likely causes such as rate limits, service overload, or context pressure from event details.

中文简介正在准备中。当前显示英语简介。

• Investigating a spike in API errors
• Finding which sessions and models are affected
• Classifying likely rate limit or overload causes

Compare one or several recent sessions with a rolling population using Agent Monitor data. See cost, total tokens, tool count, and workflow complexity as percentiles, with averages and deviation scores to identify typical sessions or outliers.

中文简介正在准备中。当前显示英语简介。

• Checking whether a session's cost was typical
• Finding unusually token-heavy sessions
• Comparing tool use and workflow complexity

Set a Claude Code spending budget by checking current costs and translating a dollar limit into a token threshold. Review existing guardrails and the proposed alert fields, then create a cost alert rule in Agent Monitor after confirmation.

中文简介正在准备中。当前显示英语简介。

• Set a monthly or weekly spending limit
• Turn a dollar budget into a token threshold
• Review existing cost alert rules

Analyze Claude Code prompt-cache usage, reuse, and costs from Agent Monitor data. See whether cached context is being read often enough to offset cache writes, identify sessions with poor reuse, and review where caching may be costing more than it saves.

中文简介正在准备中。当前显示英语简介。

• Checking whether prompt caching is reducing spend
• Finding sessions with more cache writes than reads
• Comparing cache read and write costs