luigidemasi opened a new pull request, #27850: URL: https://github.com/apache/camel/pull/27850
# Description Implements [CAMEL-25669](https://issues.apache.org/jira/browse/CAMEL-25669) with an explicit `camel-semantic-openai` expert backed by the native OpenAI Decisions API. The expert assesses the selected input; Camel routes retain control of the action taken from that assessment. - Add native Decisions support to `camel-openai` and upgrade the OpenAI Java SDK to 4.78.0. The new bridge depends on `camel-semantic` and `camel-openai`; `camel-openai` stays independent of Semantic. Require an explicitly configured Decisions model. - Preserve Semantic Boolean, choice String and score Number results. Scores retain the fractional, zero-based rubric index; provider probabilities and optional confidence remain in result metadata. Boolean uses a default threshold of 0.5, an optional override and an optional uncertainty band. Refusals, invalid provider responses and service failures have dedicated exceptions. - Support text, images and ordered mixed input. Image bodies accept byte arrays, streams and base64 strings with an image Content-Type, or self-describing image data URLs. Validate every expert's input before batch inference; TypeSafe AI keeps text/structured input and Wolf Defender keeps text input. Audio, hosted image URLs and file IDs are unsupported. - Label image input in audit output even without payload capture. Add bounded inline CLI/TUI previews, explicit image-file evaluation and audit export, while keeping unchanged captured text literal in the playground. Default display output omits image base64; full export is explicit and does not overwrite files. Target release: **4.24.0**, with no 4.23 backport planned. Migration notes and generated metadata/catalog copies are included. The TypeSafe AI and Wolf Defender module renames remain in their separate issue. The four signed and signed-off commits separate native Decisions support, Semantic media/audit contracts, the new expert module, and JBang CLI/TUI support. # Validation - 3,303 tests passed across Semantic (427), OpenAI (364), TypeSafe AI (202), Wolf Defender (34), Semantic OpenAI (14), JBang core Semantic suites (132), and the full TUI suite (2,130 passed, one existing skip). - Source checks passed for all seven affected modules. - Root reactor: `mvn -B -ntp -T 2 clean install -DskipTests -DskipITs` passed for all 704 modules; generated changes are committed. - Documentation validated with Antora for each commit. - Two opt-in Decisions API smoke tests compile and skip when disabled. Live API calls and a manual terminal TUI session were not run. # Target - [x] I checked that the commits target the correct branch (`main`). # Tracking - [x] The change has a linked JIRA issue: CAMEL-25669. # Apache Camel coding standards and style - [x] Each commit has a meaningful subject and body. - [x] I ran the full root build with tests skipped after the targeted suites and committed all generated changes. # AI-assisted contributions - [x] AI-assisted commits include co-authorship attribution, and this description identifies the tool used. _AI-assisted implementation. Generated by OpenAI Codex via /oss-fix-issue on behalf of @luigidemasi (Luigi De Masi)._ -- 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]
