lizhimins opened a new pull request, #954:
URL: https://github.com/apache/rocketmq-dashboard/pull/954

   ## Summary
   
   - Introduce an `AgentProvider` gateway abstraction with two CLI subprocess 
implementations: claude-code (`claude -p`, streaming via `--output-format 
stream-json --verbose --include-partial-messages`) and qoder (`qodercli -p`). 
Credentials reach the subprocess only through the environment (`RMQ_LLM_TOKEN` 
/ `RMQ_ANTHROPIC_BASE_URL`); the server image now ships Node.js plus both CLIs.
   - The gateway resolves the engine per request (user preference stored in the 
browser via a persisted store) falling back to the global LLM setting.
   - When `enhance` is requested, the gateway rewrites the user prompt into a 
structured RocketMQ-expert prompt first and streams the rewrite to the client 
as incremental `enhance` SSE events, rendered as a collapsible chain-of-thought 
block.
   - Home submit now sends instantly from the AI page; the home model list is 
fixed to the seven gateway models with `qwen3.8-max` recommended; the existing 
"Prompt 增强" toolbar button toggles enhancement; the home tab order becomes AI 
chat first; vitest runs with 32 workers.
   
   ## Test plan
   
   - [x] Backend tests pass; LlmConfigServiceTest / gateway tests updated
   - [x] Frontend tests pass (home/ai/studio suites), build passes
   - [x] Deployed: streaming enhance verified end-to-end (182 incremental 
enhance events for one question)


-- 
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]

Reply via email to