[docs] feat: update model support skill with encapsulation guidance

[cuichenx]

Summary

• Clarify that LLMs should use the bridge-only pattern — no provider file needed; the base GPTModelProvider returned by super().provider_bridge() is sufficient for all LLMs
• Add new "Encapsulating model-specific layers" section with three strategies for handling custom architectural blocks: local mapping subclasses, bridge hook overrides, and VLM-only custom providers
• Expand the quick decision tree with custom layer/weight layout handling branches

Test plan

• Verify skill renders correctly in Markdown
• Confirm guidance aligns with recent model additions (GLM, Nemotron-H, MiniMax-M2)

Summary by CodeRabbit

• Documentation
  • Updated guidance for implementing new AI model support with simplified LLM and VLM workflows
  • Clarified best practices for organizing model-specific configurations and layers
  • Enhanced decision tree to simplify developer implementation choices
  • Expanded with concrete strategies for model integration patterns

[coderabbitai]

📝 Walkthrough

Walkthrough

Documentation update for model support skill that removes the requirement for provider files in most LLM cases, revises VLM workflow guidance, introduces a new section on encapsulating model-specific layers, expands strategy guidance with concrete examples, and updates decision tree routing.

Changes

Cohort / File(s)

Summary

Model Support Documentation skills/adding-model-support/SKILL.md

Updated LLM file layout guidance to remove need for <model>_provider.py in most cases; revised VLM workflow to prioritize bridge/config; added new section on encapsulating model-specific layers in dedicated directories; expanded strategic guidance with local mapping subclasses, bridge hook overrides, and provider subclass criteria; updated quick decision tree with simplified LLM routing and custom layer handling.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: updating documentation for model support with new encapsulation guidance.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Test Results For Major Changes	✅ Passed	PR contains documentation-only changes to SKILL.md with no code modifications, no API changes, and no impact on functionality, numerics, or performance.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch chcui/update-model-support-skill

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

skills/adding-model-support/SKILL.md (1)

99-111: ⚠️ Potential issue | 🟡 Minor

Use a consistent modeling/modelling directory naming convention

Line 99 uses modelling_<model>/ while Line 111 uses modeling_<model>.py. If both aren’t intentionally distinct conventions, this is likely to cause copy-paste setup errors.

Suggested doc fix

-└── modelling_<model>/        # If using Megatron vision encoder
+└── modeling_<model>/         # If using Megatron vision encoder

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@skills/adding-model-support/SKILL.md` around lines 99 - 111, The docs mix
British and American spelling for the module directory/file names causing
potential copy-paste errors; pick one convention and make names consistent
(e.g., change `modelling_<model>/` to `modeling_<model>/` or vice versa) so the
directory name and the file `modeling_<model>.py` align, and update any
references in the same section to use the chosen symbol (`modeling_<model>` or
`modelling_<model>`) uniformly.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@skills/adding-model-support/SKILL.md`:
- Around line 119-120: The sentence overstates that GPTModelProvider covers "all
LLMs"; update the documentation text to clarify that super().provider_bridge()
returns a provider determined by PROVIDER_CLASS and that GPTModelProvider is
sufficient only for GPT-style LLM implementations — for non-GPT providers
authors should verify the returned provider type from provider_bridge() and
implement provider-specific logic or subclassing as needed. Reference
GPTModelProvider, provider_bridge(), and PROVIDER_CLASS in the revised sentence
to make the behavior explicit.

---

Outside diff comments:
In `@skills/adding-model-support/SKILL.md`:
- Around line 99-111: The docs mix British and American spelling for the module
directory/file names causing potential copy-paste errors; pick one convention
and make names consistent (e.g., change `modelling_<model>/` to
`modeling_<model>/` or vice versa) so the directory name and the file
`modeling_<model>.py` align, and update any references in the same section to
use the chosen symbol (`modeling_<model>` or `modelling_<model>`) uniformly.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 062513ca-81d8-4b31-8ebc-2c19d9e0f41f

📥 Commits

Reviewing files that changed from the base of the PR and between ad27e2c and c1e6ea6.

📒 Files selected for processing (1)

• skills/adding-model-support/SKILL.md