> ## Documentation Index
> Fetch the complete documentation index at: https://parser.xdr.ooo/llms.txt
> Use this file to discover all available pages before exploring further.

# ChatGPT: test the proposed fix on other logs

> Test a suggested rule against different events before approving it.

# ChatGPT's fix looks good. How do I know it works on other logs?

**Situation:** A ChatGPT answer passed **Check ChatGPT Result**, but you want to know whether it also works with different values and event variations.

## Walkthrough

1. Keep your original source and accepted preliminary review open in **ChatGPT Assisted Parser**.
2. Find **Additional ChatGPT test logs** if the candidate changed a rule or mapping and requires an independent test.
3. Paste several **different records from the same source**, not the exact sample you previously used to ask ChatGPT.
4. Select **Confirm test examples are different** only when this is true.
5. Run the offered comparison. Check that the new proposal still extracts previously correct fields and fixes the intended problem in the new examples.
6. Only if the result passes, select **Approve tested ChatGPT fix** and proceed to the generation action provided.
7. If the comparison fails, return to the correction step instead of approving.

For example, if an action value was \`allow\` in the initial sample, use separate \`deny\` and \`block\` events with different documentation-range addresses for the additional test. Vary quoted strings or optional fields when those occur in the actual source.

<img src="https://mintcdn.com/xdr-parser/DkeLTskQMJz9hCfD/images/06-chatgpt-assisted.png?fit=max&auto=format&n=DkeLTskQMJz9hCfD&q=85&s=227ecdd5895000748325a35f381a5734" alt="Parser Studio ChatGPT-assisted preparation" width="1440" height="1648" data-path="images/06-chatgpt-assisted.png" />

*The original logs stay local; share only examples you deliberately mask and approve for manual copying.*

## Expected review outcome

The revised rule works for both the original events and the additional independent examples without changing unrelated values.

## Recovery path

Do not approve a candidate that regresses a previously correct field. Ask for a narrower supported revision, then re-test.

**Related:** [Holdout validation](/workflows/holdout-validation).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.