Agent Activity Log
The Agent Activity Log page provides visibility into the agent-side activity that OX reviewed across your organization. This includes prompts developers send to their AI coding agents and tool use the agents perform, such as MCP server calls, bash commands, and external SaaS calls.
The log covers both activities that OX reviewed and approved, and activities where OX took action, such as providing inline guidance or blocking a tool call.
Reviewing this page helps you see the full picture of what VibeSec does for each agent action, including the reviews that happen in the background and never surface in the agent chat.
Activity appears in real time, so new rows stream in without a browser refresh. Showing the last few hours, older events are deleted.
Activity table
The activity table lists each agent action with a summary of how VibeSec reviewed it. An Activities counter at the top of the page shows how many rows match the current filters.
Action
The type of action the agent performed, such as User Prompt or Tool Use. Each row also shows context for the action: for a prompt, the redacted prompt notice; for a tool call, the tool name with the executed command or arguments.
When
When the agent action occurred, shown as a relative time.
User
The user who performed the action.
Role
The role of the user who performed the action in the OX platform.
AI Tool
The coding tool that produced the action.
OX Action
The outcome of VibeSec's review. See OX Action values.
OX Action values
No guidelines needed
VibeSec reviewed the prompt and no guideline applied.
Security guidelines applied
A security guideline was returned for the prompt.
Tool validated
The tool call was reviewed and no policy blocked it.
Tool blocked by policy
A policy was triggered and the tool call was blocked.
Filtering activity
The filter panel is shown by default at the left side of the page. Select Hide Filters to collapse it.
Each filter section has a search box, a checkbox for each value, and a count of matching activities next to each value. Counts update as you apply other filters.
Activity Type
Filter by the type of action, such as Prompt or Tool Use.
OX Action
Filter by the review outcome shown in the OX Action column.
Source
Filter by the AI coding tool that originated the action.
User
Filter by the user who performed the action.
Activity details
Select an activity to review how VibeSec evaluated it and what action, if any, was taken.
Depending on the activity type, OX displays either User Prompt details or Tool Use details.
User Prompt details
Select a User Prompt activity to review how VibeSec evaluated the prompt and whether security guidance was provided to the agent.
Status card
Shows the outcome of VibeSec's review.
For example, Security guidelines provided indicates that VibeSec returned security guidance to the agent.
Event Metadata
Displays technical details about the prompt event, as follows:
Model: The AI model associated with the event, when available.
Application: The application associated with the event, when available.
Event ID: A unique identifier for the event.
Tool Use details
Select a Tool Use activity to review how VibeSec evaluated the tool invocation.
Status card
Shows the outcome of VibeSec's review.
For example, Tool validated indicates that the tool invocation was reviewed against active policies and allowed to proceed.
Event Metadata
Displays technical details about the tool use event, as follows:
Model: The AI model associated with the event, when available.
Application: The application associated with the event, when available.
Tool Type: The category of tool that was invoked.Tool: The tool that was invoked.
Event ID: A unique identifier for the event.
Last updated