fix(opencode): deterministic attachment for mimo-v2.5 via _KNOWN_VISION_MODELS - #84
fix(opencode): deterministic attachment for mimo-v2.5 via _KNOWN_VISION_MODELS#84CtriXin wants to merge 1 commit into
Conversation
opencode_model_config used only OPENCODE_IMAGE_INPUT_MODELS (hardcoded duplicate) to decide whether to set attachment/modalities on a model. When that set was cleared the env-dependent capability resolver took over and the lite_pro roster test became flaky across environments. Add _KNOWN_VISION_MODELS from mms_config_web as a deterministic first-source (single source of truth per the WebUI vision config), keeping OPENCODE_IMAGE_INPUT_MODELS as a fallback. mimo-v2.5 is in _KNOWN_VISION_MODELS (vision-capable); mimo-v2.5-pro is not (text-only), matching the existing test contract. Closes #58 Agent-Model: claude-sonnet-4.5 Agent-Family: anthropic Agent-Session: 019ed982-ac79-7b71-9241-36259273559b Agent-Step: 0.1.0 Generated-By: looper 0.0.0-dev (runner=worker, agent=codex)
Worker Result
Executor 已创建/更新此 PR;下方 Gate comments 记录 review 与 merge readiness。 |
Flywheel 状态
这是 sticky status;Worker/Gate/Fix 的不可变过程记录见下方 comments。 |
AI Review Dry RunTarget
Diff Summary
Findings
Digger Policy
Project Validation
MMS Semantic Review
Inline Comment Plan
Validation Notes
PostingPR/MR comment posting is enabled. Digger will create or update one marker-based comment on reruns. Digger automatic review. |
Gate Run #1|committee
|
Gate Run #2|委员会
|
Gate Run #3|委员会
|
Summary
Fix env-dependent
KeyError: 'attachment'intest_core_opencode_lite_pro_builds_multi_model_roster(issue #58).The
opencode_model_configfunction previously relied solely on the hardcodedOPENCODE_IMAGE_INPUT_MODELSset to decide whether to setattachment/modalitieson a model config. When that set was cleared (commita46a0d6a), the fallback went through an env-dependent capability resolver that reads local snapshots and base URLs, making the lite_pro roster test flaky across environments.Fix
Add
_KNOWN_VISION_MODELSfrommms_config_web.pyas a deterministic first-source for vision capability, keepingOPENCODE_IMAGE_INPUT_MODELSas a fallback:_KNOWN_VISION_MODELSis the WebUI's authoritative vision-model set (single source of truth permms_config_web.py's vision routing config).mimo-v2.5is in_KNOWN_VISION_MODELS(multimodal);mimo-v2.5-prois not (text-only) — matching the existing test contract attests/test_opencode_launcher.py:136-142.mms_pi_supportalready imports frommms_opencode_config).Test results
test_opencode_launcher.py: 51/51 passed (1 pre-existing unrelated failure excluded).main).Closes #58
🔁 Powered by Looper · runner=worker · agent=codex · An autonomous AI dev team for your GitHub repos.