luigidemasi opened a new pull request, #27791: URL: https://github.com/apache/camel/pull/27791
Semantic audit needs to preserve what an expert returned, what that result meant at invocation time, and where the evaluation ran. This change captures that context automatically and exposes real result counts, execution timings and producer-supplied comparisons through shared monitoring snapshots. Implements [CAMEL-25618](https://issues.apache.org/jira/browse/CAMEL-25618). - Publish typed semantic AI events with immutable, complete interpretation. Each retaining consumer applies its own budget; the audit recorder creates a bounded projection before enqueueing. Captured raw input remains opt-in and outside shared events. - Remove preview route-decision recording APIs and require an explicit expert on every evaluation. Migrate Java/XML/YAML declarations, providers, schemas, examples and upgrade guidance. Persisted v1 decision records remain readable as legacy data. - Add capability-aware readers, bounded bulk export and source-local case navigation, keeping missing records, retrieval failures and unknown completeness distinct. - Share producer-owned result measurements, duration statistics, comparisons and health through the monitoring console, read-only JMX, CLI, TUI and MCP. Visualization controls do not alter collection or comparison policy. - Rework the TUI Audit and Monitoring inspectors, add expert/definition aggregation selection and bounded scrolling, and place Load params / Clear params beside Run. Explicit input inspection and draft creation do not trigger inference. The eight commits separate explicit experts, result contracts, event capture, reader/case access, monitoring runtime, CLI integration, TUI/MCP integration, and scalar parameter rendering. Generated metadata and applicable documentation travel with their changes. Validation: - 727 tests passed across the semantic runtime, TypeSafe AI, Wolf Defender, CLI connector, YAML DSL, JBang core, TUI and MCP; two existing integration tests skipped. - After rebasing onto current main, 149 TUI tests passed with TamboUI 0.6.0, including the semantic views and upstream wrapping changes. - Formatter/import checks and generated metadata refreshed; documentation checked with Antora. - The semantic runtime suite also passes at each of the first five commit stages. - Local demo checks exercised actual expert results, monitoring aggregates, parameter controls and Audit scrolling at multiple terminal sizes. - Full reactor and external integration suites were not run. The wide Audit table still assigns spare width to Result; column-space balancing remains a presentation follow-up. _AI-assisted implementation and PR description by Codex on behalf of @luigidemasi._ -- 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]
