The Log Analysis Agent delivers the most accurate and useful results when logs are provided with sufficient detail and context. Proper usage practices improve error detection, KEDB matching accuracy, and the overall quality of troubleshooting insights.
This article outlines recommended best practices for interacting effectively with the Log Analysis Agent.
How to Ask Questions Effectively
The Log Analysis Agent performs best when the message contains one of the following:
- A log path
- Pasted log content
- A follow-up question related to a prior analysis session
Recommended Best Practices
Clearly Provide the Log Source
Users should either:
- Paste the log block directly into the chat
- Provide a valid file path or storage URI
Supported examples include:
/var/log/...gs://bucket/...s3://bucket/...- Console URLs
Include the Full Stack Trace
The agent performs better when provided with:
- Exception classes
- Error messages
- Stack frames
- Nested
Caused by:blocks
Providing the complete stack trace improves KEDB matching accuracy.
Include Surrounding Context
Some failures reference earlier lines or depend on chained exceptions. Including surrounding log context helps the agent reconstruct the full failure sequence.
Continue Follow-Ups in the Same Session
The agent remembers prior analyses within the same chat session. This allows users to ask follow-up questions without re-uploading the logs.
Example follow-up questions include:
- “Explain error #2 more”
- “What would the workaround break?”
- “Can you elaborate on the root cause?”
Save Unknown Errors to the KEDB
When unknown errors are detected, users can confirm with:
Saveyes, save
This creates new KEDB entries to improve future automated analysis.
Provide Feedback on Existing Fixes
Operational feedback improves KEDB quality over time.
Examples:
- “That fix worked”
- “The workaround for KE-1234 didn't help”
Things to Avoid
Do Not Paste Screenshots
The agent cannot interpret images. Logs must be provided as text.
Do Not Truncate Errors
Providing only the final exception line reduces diagnostic accuracy.
For example:
NullPointerExceptionwithout surrounding stack traces is insufficient for proper KEDB matching.
Avoid Mixing Topics in the First Request
Users should avoid combining log analysis requests with unrelated platform questions in the same initial message.
Avoid General Platform Questions
The Log Analysis Agent is intentionally scoped to troubleshooting tasks and will redirect general platform questions to the Syntasa Help Agent.
Conclusion
Using the Log Analysis Agent effectively requires clear log inputs, complete stack traces, and focused troubleshooting questions. Following these best practices improves analysis accuracy, enhances KEDB matching, and enables faster operational resolution.