> ## Documentation Index
> Fetch the complete documentation index at: https://docs.caylex.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Analytics & Session Logs
> Monitor tool usage, performance, errors, and user activity across your deployments.
Caylex provides real-time analytics and detailed session logs for every interaction that passes through the Navigator. Use these to monitor performance, debug issues, and understand how your tools are being used.
## Dashboard overview
The analytics dashboard is accessible from the main navigation in the Caylex platform. It provides a high-level view of your platform's activity with the following metrics:
### Key metrics
| Metric | Description |
| -------------------------- | ----------------------------------------------------------------------- |
| **Sessions processed** | Total number of sessions processed |
| **Tool calls** | Total number of individual tool calls executed |
| **Error rate** | Percentage of tool calls that resulted in errors |
| **Average latency** | Average time for the Navigator to process a tool call |
| **Average runtime** | Average execution time of tool calls on external servers |
| **Active users** | Weekly active users (WAU) and monthly active users (MAU) |
| **Tool calls per session** | Average number of tool calls per session (indicates session complexity) |
All metrics include **period-over-period comparisons** — you can see how the current period compares to the previous one (e.g., this week vs last week) to identify trends.
### Time-series graphs
The dashboard includes interactive graphs for:
* **Sessions over time** — visualize session volume trends
* **Tool calls over time** — track tool execution volume
* **Error rate over time** — monitor error trends and spikes
* **Error types over time** — see the breakdown of error categories (authentication, rate limit, timeout, execution errors)
* **Latency over time** — track Navigator processing time trends
* **Runtime over time** — monitor external server response times
* **Tool calls per session over time** — observe session complexity trends
Use the time window selector to view data over different periods: last hour, last day, last week, or last month. You can also adjust the aggregation frequency (per minute, hour, day, week, or month) for more or less granularity.
### Top tools and servers
The dashboard also shows:
* **Top tools** — the most frequently used tools, ranked by call count
* **Servers** — which servers are being accessed and how often
* **Navigator-tool connections** — which navigators are using which tools, visualized as a relationship map
## Filtering
All analytics can be filtered by:
| Filter | Description |
| --------------- | -------------------------------------- |
| **Project** | Scope to a specific project |
| **Navigator** | Scope to a specific navigator instance |
| **Server** | Scope to a specific server instance |
| **User** | Scope to a specific user |
| **Time window** | Select the date range to analyze |
Combine filters to drill down into specific areas. For example, filter by a specific navigator and server to see how that navigator uses a particular integration.
## Session logs
Session logs provide a detailed view of every session processed by the Navigator. Navigate to a project and open the **Session Logs** tab to see all sessions.
Each session log entry shows:
| Column | Description |
| ------------------- | ----------------------------------------------------- |
| **Session Summary** | The original query or intent that started the session |
| **Tool Calls** | Total number of tool calls executed in the session |
| **Errors** | Number of tool calls that failed |
| **Servers Invoked** | Icons showing which servers were accessed |
| **Navigator** | Which navigator instance processed the session |
| **User** | Which user triggered the session |
| **Timestamp** | When the session was received |
You can filter sessions by navigator, server, and date range. Sort by newest or oldest first.
### Session details
Click on any session to open the **Session Details** drawer with three sections:
#### Session summary
The full text of the original query or intent.
#### Statistics
An overview of the session including the user, total tool calls, error count, timestamp, and which servers were invoked.
#### Tool trace
A visual timeline showing every tool call executed in the session, displayed as a vertical step-by-step trace. The trace shows:
* **Individual tool calls** — each displayed with the tool name, timestamp, and server icon
* **Parallel tool calls** — grouped together and labeled as "Parallel" with the number of tools executed simultaneously
* **Error indicators** — failed tool calls are highlighted with a red icon
For each tool call in the trace, you can see:
| Field | Description |
| --------------- | ------------------------------------ |
| **Tool** | The name of the tool that was called |
| **From Server** | Which server the tool belongs to |
| **Latency** | How long the call took |
| **Timestamp** | When the call was executed |
If a tool call failed, you can click **View Error Message** to see the full error details, including any contextual error recovery information with examples of previous successful calls to the same tool.
Session traces are invaluable for debugging. If an agent is not behaving as expected, the trace shows exactly which tools were called, on which servers, and in what order — making it easy to pinpoint where things went wrong.
## User analytics
Per-user analytics are available in the project's **Users** tab. For each user, you can see:
* **Activity timeline** — sessions and tool calls over time
* **Top tools** — the tools used most by this user's interactions
* **Top servers** — the servers accessed most on behalf of this user
* **Error rate** — the percentage of tool calls that failed for this user
See [Managing Users](/auth/managing-users) for more on the user view.
## Efficiency analytics
The navigator instance analytics drawer includes two metrics that measure how effectively your agent uses the Navigator's tool suggestion system. These appear under **Efficiency Analytics** on the navigator instance's Analytics tab.
| Metric | Description |
| ----------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Tool Discovery Efficiency** | The ratio of tool invocations to tool discovery operations (`suggest_tools` + `get_tool_schemas` calls). Higher values mean your agent is invoking more tools per discovery operation — indicating efficient tool selection with less overhead. |
| **Tool Suggestion Spread** | The ratio of tool invocations to the total number of suggested tools. Higher values mean a larger proportion of suggested tools are actually being invoked, indicating that the suggestions are relevant and high quality. |
These metrics are scoped to the selected time window (hour, day, week, or month) and are filtered by project and navigator instance.
## Error monitoring
The error analytics section categorizes errors by type:
| Error type | Description |
| ------------------ | ---------------------------------------------------------- |
| **Authentication** | Credential issues — user not authenticated, expired tokens |
| **Rate limit** | External server rate limiting |
| **Timeout** | Tool call exceeded the configured timeout |
| **Execution** | Errors returned by the external MCP server |
| **Query** | Errors in the query processing itself |
Use error type trends to identify systemic issues. For example, a spike in authentication errors might indicate expired OAuth tokens that need renewal, while timeout errors might suggest a server performance problem.