Execution & Monitoring
One of Agent HQ’s most powerful features is the ability to watch AI agents work in real time. When a task is processing, you get full visibility into what’s happening.
Task Detail view
Section titled “Task Detail view”Click any task card on the board to open the Task Detail modal. For running tasks, this shows live execution data.
Execution logs
Section titled “Execution logs”Logs stream in real time as the task executes. Each log entry has a type:
| Type | Description |
|---|---|
| system | System messages (setup, teardown) |
| text | Standard output |
| tool | Tool invocations (file writes, commands) |
| error | Errors and warnings |
| output | Final output and results |
Logs are displayed in a terminal-style view with color coding.
File viewer
Section titled “File viewer”As Pilot writes code, modified and created files appear in a file tree. Click any file to see its contents with syntax highlighting.
For files that Pilot annotated (inline comments or drawing annotations), those annotations appear alongside the code.
Test results
Section titled “Test results”If the task runs tests, results are parsed and displayed in a structured format showing:
- Number of tests passed / failed / skipped
- Individual test names and their status
- Error messages for failed tests
Web preview
Section titled “Web preview”For tasks that produce a web application or page, a live preview is available in an iframe. This lets you see the visual result without leaving Agent HQ.
Artifact output
Section titled “Artifact output”Tasks can produce markdown artifacts — structured output like documentation, reports, or plans. These render with full markdown formatting in the task detail view.
Plan view
Section titled “Plan view”For complex tasks, Pilot may create an execution plan — a series of steps to accomplish the goal. The plan view shows each step with its status (pending, in progress, done, failed).
Sandbox details
Section titled “Sandbox details”Each task runs in an isolated sandbox environment that includes:
- Filesystem — an isolated copy of the project’s repo
- Terminal — shell access for running commands
- Browser — Puppeteer-based browser for screenshots and web automation
Task completion
Section titled “Task completion”When a task finishes, it moves to the “Done” column. The task detail shows:
- Total execution time
- Token usage (input and output)
- Cost in dollars
- All generated files and changes
Retrying failed tasks
Section titled “Retrying failed tasks”If a task fails, you can:
- Inspect the logs — understand what went wrong
- Modify the task — edit the title or body to provide better instructions
- Retry — click the retry button to re-run the task from scratch
Screenshots
Section titled “Screenshots”You can capture browser screenshots of a task’s sandbox at any point. Screenshots are stored and viewable in the task detail, and can be annotated with inline comments or drawings.