CaroFernando commented on issue #8682: URL: https://github.com/apache/texera/issues/8682#issuecomment-5881945089
Since the text was copied from the chat bubble, my guess is the JSON was part of the agent's message itself: it looks like the payload the agent uses internally to modify the workflow (applied automatically, not meant for manual use), just surfaced as a code block, which is understandably confusing. The review-before-apply idea from earlier in this thread seems like the right direction, and might address both halves at once: proposed edits could arrive in the chat as an Approve/Reject card showing a plain-language before and after summary of the workflow, so there's no raw JSON to interpret, and nothing lands until approved. Read-only actions (viewing operators, checking schemas) would probably still run automatically, so it's not an approval prompt every step. It would touch both the agent service and the UI, so it's likely a staged series rather than one PR. @bobbai00, Ma77Ball mentioned past experiments in this direction; is there existing work to build on? @hhvu0102, one check that could confirm this diagnosis: the Export button in the chat toolbar (top right) dumps the steps as JSON; the content field of the step that was copied from would tell us what that block actually is. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
