davsclaus commented on PR #27602: URL: https://github.com/apache/camel/pull/27602#issuecomment-6076672854
A few more TUI points, on top of the review above. The F1 help (`semantic.md`) and the adoc pages look good, thanks. **AI/MCP visibility of the Semantic screen** The TUI MCP tools (`tui_get_table` etc.) let an AI see and drive what the user sees. `SemanticTab.getTableDataAsJson()` currently returns only the metadata rows (view, rows, selected index, selected definition and expert). Could it also include: - the selected expert **operation** and the focused pane - the **Try expert** state: draft input, parameter values, and the last result (value, probability/probabilities, confidence, elapsed time) or error - the same for the **sample evaluation** popup when it is open Also, `SemanticTab` doesn't implement `setInputValue`, which the other input screens do (e.g. SQL Query, HTTP). Without it, an AI driving the TUI can't fill in the input/parameters and run an expert, even though a person can. Supporting it for the playground and sample fields would make the screen usable from the MCP side too. **Follow-up: CLI command** I opened [CAMEL-25489](https://issues.apache.org/jira/browse/CAMEL-25489) for Camel CLI commands that use the same `semantic-metadata` / `semantic-evaluate` actions (list definitions/experts, evaluate a definition or an expert operation). It's not needed for this PR. It would be one more client for `semantic-evaluate`, though, so the connector-thread concern from the review matters there as well. _Claude Code on behalf of davsclaus_ _This comment was generated by an AI agent and may contain inaccuracies. Please verify all suggestions before applying._ -- 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]
