Activity Dashboard

The app ships with an AI Agent Activity dashboard, giving administrators an operational view of how the agent is being used across the organization.

Open it from the Dashboards menu in the Lynx AI Agent app.

Note

The dashboard is available to administrators only. It reads from the _internal index, which the lynx-ai role does not grant on its own - regular users see no results.

Panels

The dashboard is filtered by time range, user, and model, and covers:

Volume and reliability - total requests, success rate, errors, unique users, and average request duration.
Activity timeline - requests, tokens, and average duration over time.
Models - request distribution and token consumption per model, including reasoning tokens.
Tools - the tools the agent invoked most, with their success rates.
Skills - skill activations and success rates, in total and over time.
Connectors - connector calls and success rates, in total and over time.
Errors - the most frequent error types and their distribution over time.

Data Source

The app writes a usage event to Splunk's _internal index for each request. The dashboard reads them with:

index=_internal source=lynx_ai_agent sourcetype=lynx_ai_agent:usage

Events record the user, model, tools, skills, token counts, duration, and outcome of the request. Prompt and response content is not included.

Each event carries an event field naming what it describes:

`event`	Written when
`request`	A generation finishes. Carries the token counts and duration.
`tool`	The agent invokes a built-in or connector tool.
`skill`	The agent loads a skill.
`dashboard_save`	A dashboard is saved or a preview fails to save.
`context_trim`	A conversation had to be trimmed to fit the model's context window.

Every event also carries a status (success, error, or cancelled); failures add an error_type and a short error_message, which is what the error panels group by.

Logs

Alongside the usage events, the app writes operational logs to _internal under the same source:

index=_internal source=lynx_ai_agent sourcetype=lynx_ai_agent:log

`sourcetype`	Contains
`lynx_ai_agent:log`	Operational and agent-behavior messages, with a severity, a description, and the chat ID and model that were in play.
`lynx_ai_agent:user`	Changes users make to their own settings.
`lynx_ai_agent:usage`	The usage events behind the dashboard above.

These logs are the first place to look when something misbehaves, and are worth including in a support request.