knowledge: daily intake batch 18

[tzhouam]

Daily knowledge-intake batch from omni-reviewbot: executable rules distilled from merged PRs and copilot bugfix runs, append-only per the knowledge contract. Merging this PR is the human promotion gate — review the rules like any other knowledge edit.

Proposed rules

• knowledge/repos/vllm-omni/components/diffusion/rules-system-runtime.md → DIFF-4ad (sources: PR #7548)
• knowledge/repos/vllm-omni/models/minicpm-o-4-5/rules-cuda-graphs.md → MCPMO-1k (sources: PR #7416)
• knowledge/repos/vllm-omni/models/minicpm-o-4-5/rules-duplex.md → MCPMO-4j (sources: PR #7416)
• knowledge/repos/vllm-omni/benchmark/rules.md → BENCH-1f (sources: PR #7624)
• knowledge/repos/vllm-omni/models/qwen-omni/rules.md → QOMNI-1h (sources: PR #7304)
• knowledge/repos/vllm-omni/models/cosyvoice3/rules.md → COSYVOICE3-1c (sources: PR #7521)
• knowledge/repos/vllm-omni/models/qwen-image/rules.md → QWENIMG-1d (sources: PR #7513)

Rejected proposals (not applied)

• proposal 3 (attempt 0): rule_id already exists on knowledge/repos/vllm-omni/components/serving/rules-engine-lifecycle.md

Source events

• merged_pr PR #7174: [Diffusion] MAGI-2: regional compile coverage
• merged_pr PR #7548: [Bugfix] Skip generic text-only warmup for DreamZero
• merged_pr PR #6845: [LingBot World] Reuse the text encode for a prompt already encoded
• merged_pr PR #7413: [Core][Frontend] Unified Full-duplex Framework
• merged_pr PR #7535: [Attention] [P1] Keep H3 VSA policy in the model and share sparse primitives
• merged_pr PR #7416: [Bugfix][MiniCPM-o] Stabilize code2wav CFM CUDA graph caching with reference audio normalization and mel-frame bucketing
• merged_pr PR #7483: [Frontend] Add MiniMax-H3 references and Ref2VA workflow
• merged_pr PR #6939: [feat] Extend disaggregated MiniMax-H3 stage with VAE encoders
• merged_pr PR #7423: [Frontend] Add MiniMax H3 ComfyUI text-to-video workflow (WF-01)
• merged_pr PR #6649: [Boogu-Image] Fuse QKV/FFN projections + switch to diffusion RMSNorm
• merged_pr PR #5921: [Diffusion] Fuse Qwen-Image-Edit select01 modulation for CUDA
• merged_pr PR #7331: [Bugfix][Omni-DuplexEval] Load local Hugging Face dataset layouts in duplex eval loader
• merged_pr PR #7277: [Bugfix] Fail a duplex session that outgrows max_model_len instead of killing the EngineCore
• merged_pr PR #7521: [Perf][CosyVoice3] Bounded-window incremental HiFT vocoder streaming
• merged_pr PR #7624: [Bugfix] Preserve benchmark metric sample counts in results
• merged_pr PR #7513: [Bugfix][Qwen-Image] Restore RotaryEmbedding CUDA RoPE for Diffusers e2e
• merged_pr PR #7657: [CI][MiniCPM-o] Lower the Daily-Omni accuracy gate to 0.77
• merged_pr PR #7092: [Bugfix] Reject failed diffusion benchmark warmups
• merged_pr PR #7454: [Misc] Remove dead deploy keys from personaplex.yaml
• merged_pr PR #7304: [Bugfix][Qwen3-Omni] Include deferred residual in captured Thinker states

Public SDK curation result

{
  "batch_id": "sha256:d0a6576f5213a7ed8204d5645dcaeffee0906a288b548369d20c6baee127003b",
  "success": true,
  "attempted": 8,
  "applied": 7,
  "accepted_indexes": [
    0,
    1,
    2,
    4,
    5,
    6,
    7
  ],
  "rejected_indexes": [
    3
  ],
  "updated_document_ids": [
    "knowledge/repos/vllm-omni/benchmark/rules.md",
    "knowledge/repos/vllm-omni/components/diffusion/rules-system-runtime.md",
    "knowledge/repos/vllm-omni/models/cosyvoice3/rules.md",
    "knowledge/repos/vllm-omni/models/minicpm-o-4-5/rules-cuda-graphs.md",
    "knowledge/repos/vllm-omni/models/minicpm-o-4-5/rules-duplex.md",
    "knowledge/repos/vllm-omni/models/qwen-image/rules.md",
    "knowledge/repos/vllm-omni/models/qwen-omni/rules.md"
  ],
  "updated_on": "2026-09-18",
  "validators": [
    {
      "validator_id": "knowledge/tools/check_knowledge_tree.py",
      "passed": true,
      "status": "passed",
      "returncode": 0,
      "output": "提醒：文件接近拆分线：general/review/guides/review-execution-contract.md (183 个非空行，21548 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/ci/rules.md (243 个非空行，27775 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/configuration/rules.md (174 个非空行，30403 bytes)\n提醒：目录有 15 个普通页面，应该考虑分类：repos/vllm-omni/components/configuration\n提醒：文件接近拆分线：repos/vllm-omni/components/diffusion/rules-attention.md (190 个非空行，25118 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/diffusion/rules-checkpoint-loading.md (211 个非空行，26702 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/diffusion/rules-output-lifecycle.md (147 个非空行，21047 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/diffusion/rules-system-runtime.md (194 个非空行，28137 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/diffusion/rules.md (172 个非空行，27612 bytes)\n提醒：目录有 22 个普通页面，应该考虑分类：repos/vllm-omni/components/diffusion\n提醒：文件接近拆分线：repos/vllm-omni/components/model-executor/rules-bridge-batch.md (134 个非空行，18619 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/model-executor/rules.md (193 个非空行，31439 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/scheduler/rules.md (299 个非空行，32496 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/serving/rules-engine-lifecycle.md (246 个非空行，32271 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/components/serving/rules.md (196 个非空行，31197 bytes)\n提醒：目录有 12 个普通页面，应该考虑分类：repos/vllm-omni/components/serving\n提醒：文件接近拆分线：repos/vllm-omni/models/hunyuan-image3/rules.md (137 个非空行，25387 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/models/minicpm-o-4-5/rules-duplex.md (102 个非空行，16465 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/models/minicpm-o-4-5/rules.md (258 个非空行，30441 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/models/minimax-h3/rules-cache-task.md (204 个非空行，28481 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/models/minimax-h3/rules-deployment.md (236 个非空行，27707 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/models/minimax-h3/rules.md (183 个非空行，32421 bytes)\n提醒：文件接近拆分线：repos/vllm-omni/models/qwen3-tts/rules.md (190 个非空行，20981 bytes)\n知识目录检查通过：0 个错误，23 个提醒"
    },
    {
      "validator_id": "knowledge/tools/check_wiki_lint.py",
      "passed": true,
      "status": "passed",
      "returncode": 0,
      "output": "wiki lint 通过：0 个错误，0 个提醒（共 388 页）"
    }
  ],
  "rolled_back": false
}

The vllm-omni release audit runs in CI on this PR as usual.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.