# Debugging failures

> How to investigate and fix a failure, in the dashboard or via the MCP server.

Rendered version: https://docs.semaloop.com/test-results/debugging-failures/

When an agent finds something wrong in your application, Semaloop already contains all the necessary context to help suggest a fix. If you want to investigate manually, you can click through the trace, and also watch the playback video to see exactly what happened.

Within the Semaloop dashboard, we also provide a failure reason (root cause analysis), that also suggests a fix. This is because our agent also has access to the system logs and network logs, which allows us to better understand where things went wrong.

Note

Both system logs and network logs can be downloaded from the Semaloop dashboard.

Additionally, our [MCP server](/integrations/mcp-server/) can be used to ingest all of the relevant context into a coding agent, which can then be used to resolve the bug. You simply need to grab the test run ID, which can be found in the Send to agent popup. For example: “Help me fix Semaloop test run 028ef19d-7e3a-4517-920f-edeac215a082” could yields a simple code fix as a coding agent has all context available to it.

## Related

- Index of every Semaloop docs page: https://docs.semaloop.com/llms.txt
- All Semaloop docs in one file: https://docs.semaloop.com/llms-full.txt