wenjin272 opened a new issue, #1249: URL: https://github.com/apache/flink-agents/issues/1249
Parent: #1232 ### Problem and scope Chat with Models needs a clear progression from basic messages and the request/response flow to multimodal content and tool calling. Framework message contracts must be distinguished from provider/model capabilities. #### Scope - [ ] Introduce ChatMessage, roles, plain-text factories, text/getText(), and relevant fields before advanced content/media blocks. - [ ] Explain the current framework block types, source kinds, media_type, and serialization contract, including the absence of a framework-wide media-format allowlist. Document bytes factories according to the delivered API. - [ ] Put a visible capability warning before the first multimodal example: representable content does not imply provider/model support. Verify the support matrix for 0.4, including the distinction between OpenAI Chat Completions and Responses. - [ ] Consolidate framework-wide multimodal explanations here. Coordinate moving them out of OpenAI-specific sections, where conversion details and provider/model limits belong. - [ ] Keep the request/response flow, examples, success/failure handling, and tool-calling introduction coherent with the message explanation. Link to Tool Use for tool declaration, parameter injection, and parallel-call behavior. #### Coordination and acceptance Build on the Chat with Models page delivered by #1244; retain completed improvements and address remaining gaps. Coordinate provider capability details with the Chat Models/provider child of #1232 and tool behavior with its Tool Use child. Related API work: #1195 (bytes factories). Verify representative snippets, capability warnings, and reciprocal links against the delivered 0.4 behavior. #### Review coverage DOC-043, DOC-044, DOC-045, DOC-046. This issue owns the framework/provider separation findings; coordinate provider-side edits without duplicating the shared explanation. -- 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]
