Skip to content

Analysis Dashboard

In Agent Monitoring > Analysis Dashboard, you can view metric information for a selected application in chart form, gaining a comprehensive understanding of application performance and cost distribution.

You can also access the Analysis Dashboard from the application entry on the Agent Overview page. When entering through a specific Agent, the system automatically applies filtering conditions for that Agent.

The top of the page supports switching between Agent / LLM state, defaulting to Agent state.

Note

When there is no data in the selected time range, the chart area will display a "No data" status. You can adjust the viewing range using the time widget in the upper right corner.

LLM Monitoring

After selecting an LLM Monitoring application, the Analysis Dashboard displays the following content.

Core Metrics Overview

The top of the dashboard shows key metric statistics for the selected time range:

  1. Request Count: Total number of requests
  2. Span Count: Total number of associated Span calls
  3. Request Error Rate: Percentage of failed requests out of total requests
  4. Total Token Consumption: Total tokens consumed by all requests
  5. Average Response Duration: Average processing time across all requests

Trend Analysis Charts

Using multi-dimensional trend charts, you can track changes in application performance and usage trends:

  1. Request Trend Chart: View request volume distribution over different time periods to identify usage peaks
  2. Request Error Rate: Shows the error rate trend, enabling quick detection of abnormal fluctuations
  3. Request Response Time: Displays Avg, P75, P90, and P99 percentile durations to identify long-tail latency issues
  4. Token Consumption Trend Chart: Shows the time distribution of token usage, aiding cost forecasting
  5. Average Time to First Token (TTFT): Measures first-token latency, reflecting model response speed

Model and User Analysis

Analyze resource usage and performance from both model and user dimensions:

  1. Request Count by Model: Donut chart showing call distribution across different models (e.g., qwen3:1.7b)
  2. Token Usage by Model: Rank resource consumption across models
  3. Request Response Analysis by Model (Top 20): Table showing detailed performance metrics for each model, including request count, average/max duration, and percentile durations
  4. User Request Count Ranking (Top 10): Identify high-frequency users
  5. User Token Consumption Ranking (Top 10): Locate high-cost users to assist cost allocation

Agent Monitoring

After selecting an Agent Monitoring application, the Analysis Dashboard displays Agent Runtime call, token, and duration data in the following sections: Runtime Overview, Model and Tokens, Tool Module, and Skill Module.

Filter Conditions

The page supports filtering dashboard data by the following conditions:

  • Agent Type
  • Agent ID
  • Agent Name
  • Model Provider

The available values for Agent ID and Agent Name update dynamically based on the parent filter selection. When you change the time range or filter conditions, the overview, distribution, trend, and table data in the dashboard refresh simultaneously.

Runtime Overview

The Runtime Overview shows the overall call and token usage within the current time range:

  • Total Operations: Sum of Tool, Skill, and LLM call counts
  • Tool Call Count: Number of Tool operation calls
  • LLM Call Count: Number of model conversation calls; a single business request may involve multiple LLM calls
  • Skill Call Count: Number of Skill calls; displays 0 when no Skill data is reported
  • Total Tokens: Sum of usage across all token types (input, output, etc.)
  • Average Tokens per LLM Call: Ratio of total tokens to total LLM calls
  • Model Provider Distribution: Proportion of LLM calls by provider
  • Model Call Distribution: Number of LLM calls by model
  • Tool Call Distribution: Proportion of calls by Tool name
  • Top 10 Skills by Call Count: Shows the 10 Skills with the highest call counts

Overview cards and distribution charts are calculated based on the counter increment within the current time range, displaying the actual new calls or token usage added during that period, rather than the cumulative counter value.

Model and Tokens

The Model and Tokens module is used to analyze model call volume, token consumption, and call duration:

  • LLM Call Trend: Shows LLM call volume trends by model
  • Input/Output Token Trend: Shows usage trends by token type
  • Token Trend by Model: Shows token usage trends by model
  • LLM Call Duration Percentile Table: Displays LLM call counts and P50, P75, P90, and P99 call durations by Agent type

Trend charts use the dashboard's automatic time interval to calculate incremental data within each period.

Tool Module

The Tool Module shows the number of active Tools, call trends, durations, and errors:

  • Active Tools: Number of Tools with call count > 0 in the current time range
  • Max Tool Duration: Maximum Tool call duration in the current time range
  • Tool Call Trend: Shows call volume trends by Tool name
  • Average Tool Duration: Shows average call duration per time interval by Tool name
  • Top 10 Tool Errors: Shows the 10 Tools with the highest error counts and their trend

Skill Module

The Skill Module shows the number of active Skills, call trends, durations, and errors:

  • Active Skills: Number of Skills with call count > 0 in the current time range; displays 0 when no Skill data is reported
  • Max Skill Duration: Maximum Skill call duration in the current time range; displays "No data" when there are no Skill calls
  • Skill Call Trend: Shows call volume trends by Skill name
  • Average Skill Duration: Shows average call duration per time interval by Skill name
  • Top 10 Skill Errors: Shows the 10 Skills with the highest error counts and their trend

Further Reading

You may also be interested in the following: