DiagnosticsPreview

Diagnostics

The Server continuously collects console output, JavaScript exceptions, failed network requests, performance timings, health information, and process status for each workspace.

Product crash reporting for Agent-Up itself is separate.

Note

Console output is the first diagnostic source after a failed start, health check, or browser timeout. The Server keeps those lines intact: one source line per entry, stdout and stderr identity on the entry, and an explicit [truncated] marker when older output is dropped.

1. Reproduce the failure in the running workspace

2. Read the application console

3. Inspect health, logs, and audit history

Command

agent-up diagnostics

Where it differs

CLI

agent-up diagnostics shows process state plus the readiness signal in use (port, health-check URL, process, or container), recent application logs, and relevant JavaScript, network, and browser errors. Entries are marked active or resolved. agent-up review-snapshot writes a JSON snapshot of the current workspace for someone reviewing an agent task.

MCP

Call Orchestration MCP get_workspace_console first after browser failures. If that tool is unavailable, query Audit MCP for recent application events from process. export_workspace_review_snapshot is a shareable snapshot after an agent task, not a substitute for the live workspace.

Next in this slice

Verification

Required checks before enqueue.