youngkermit8-coder opened a new issue, #1731:
URL: https://github.com/apache/rocketmq-dashboard/issues/1731
## Problem
`LlmConfigService.testConfig(...)` always probes
`OpenAiCompatibleLlmClient.listModels(...)` whenever the selected provider is
HTTP-compatible. It does not consider the selected engine.
For `claude-code` and `qoder`, production requests execute a local CLI
through `AgentProviderRegistry`, but the settings "test connection" action
probes an unrelated HTTP `/models` endpoint instead. A healthy installed CLI
can therefore be reported as unavailable because the HTTP endpoint or API key
is unusable, while a missing CLI can be reported as successful when the HTTP
endpoint responds.
## Reproduction
On current `rocketmq-studio` (`c2488ccb45b7ae66a3a37a6e4fab8a62a02367f4`),
configure a `claude-code` engine and make the OpenAI-compatible model endpoint
fail. A focused regression expecting the CLI path to avoid `listModels(...)`
fails deterministically:
```text
Tests run: 1, Failures: 1, Errors: 0, Skipped: 0
expected: 0
but was: 1
```
## Expected behavior
- HTTP engines continue validating connectivity through provider model
discovery.
- CLI engines test the selected `AgentProvider` availability and never call
the HTTP model endpoint.
- A missing CLI returns the existing `llm.provider.cli_missing` failure code
and an actionable hint.
- Add regression coverage for available and unavailable CLI engines, while
preserving the existing HTTP probe tests.
## Duplicate search
Searched open and closed issues and pull requests for combinations of
`LlmConfigService`, `config test`, `claude-code`, `qoder`, `AgentProvider`,
`CLI availability`, and `listModels`. No existing contribution covers
engine-aware configuration testing.
--
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]
