feat(llm_minimax): add MiniMax M3 model profile

[mithileshgau]

Summary

Adds the `MiniMax-M3` cloud profile to the `llm_minimax` node — MiniMax's frontier multimodal coding model (1M context, 128K recommended output). Includes `sync_models.config.json` override blocks so the next weekly cron's auto-PR doesn't silently downgrade M2.x or fall back to the 16K hardcoded estimate for M3.

What changed

• `nodes/src/nodes/llm_minimax/services.json` — M3 profile + field def + conditional entry; M2.x outputs upgraded 8192 → 65536 per docs-recommended; sync now sources the values from `sync_models.config.json` (visible as inline annotation comments).
• `tools/src/sync_models.config.json` — `token_limit_overrides` and `model_output_tokens.overrides` blocks populated for all 8 MiniMax cloud models, anchored to the official API reference. `protected_profiles` extended to include `minimax-m3`.
• `nodes/src/nodes/llm_minimax/README.md` — M3 row + summary in the Cloud profile table.

Test plan

• `python3 tools/src/sync_models.py --provider llm_minimax --enable-discovery` (dry-run) shows `(no changes — 20 profiles unchanged)` — the sync sees services.json as fully aligned with the overrides
• Live-test M3 in canvas using `ROCKETRIDE_MINIMAX_KEY` (verified locally on engine sync)
• CI sync cron's next run produces no spurious diff against this branch

Summary by CodeRabbit

• New Features

  • Added MiniMax M3 profile with 1M-token context; multimodal model noted and text path exposed; recommended output guidance up to 512K (typical ~128K).

• Documentation

  • Cloud catalogue and profile documentation updated to include MiniMax M3 and refreshed descriptions.

• Chores

  • Expanded provider profiles, editable MiniMax settings, and explicit per-model context & output token-limit overrides.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: db4c6b25-b8b4-4288-b982-765be4db9b81

📥 Commits

Reviewing files that changed from the base of the PR and between 90581fd and 6f67be7.

📒 Files selected for processing (1)

• tools/src/sync_models.config.json

---

📝 Walkthrough

Walkthrough

Adds MiniMax M3 support: documents the model in the llm_minimax README, adds/updates minimax-m3 and provider-based profiles plus expanded fields schema in services.json, and enables MiniMax token/context and output-token overrides in sync_models.config.json.

Changes

MiniMax M3 Support

Layer / File(s)	Summary
M3 Token Limit Configuration tools/src/sync_models.config.json	Adds minimax-m3 to protected_profiles, activates token_limit_overrides for MiniMax model IDs, and sets model_output_tokens.overrides with a higher limit for MiniMax-M3 and shared lower limits for other MiniMax variants.
M3 Profile Configuration and Schema nodes/src/nodes/llm_minimax/services.json	Updated cloud preconfig.profiles to modelSource: "provider" for cloud MiniMax models, added/updated minimax-m3 provider profile with model name and token limits, kept local *-local profiles as manual, and expanded fields schema to include editable minimax.custom properties and a minimax.minimax-m3 object.
M3 Documentation nodes/src/nodes/llm_minimax/README.md	Added MiniMax M3 row to Cloud profiles table (1M-token context) and expanded cloud catalogue text with M3 multimodal/coding notes, recommended output limits, and a clarification that the node exposes only the text path.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related issues

• Add MiniMax M3 model to llm_minimax node #1045: Directly related—adds MiniMax-M3 entries and token-limit overrides in sync_models.config.json and a minimax-m3 profile in services.json.

Possibly related PRs

• rocketride-org/rocketride-server#651: Introduced provider sync tooling that this PR's minimax provider/profile updates integrate with.
• rocketride-org/rocketride-server#967: Earlier llm_minimax profile and sync_models changes this PR builds upon.

Suggested reviewers

• jmaionchi
• stepmikhaylov
• Rod-Christensen
• asclearuc

Poem

🐰 I hop through configs, lines in tow,
M3 arrives with a million-glow,
Tokens set and profiles new,
Docs whisper what the node can do,
I nibble bytes and cheer—huzzah!

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely summarizes the primary change: adding the MiniMax M3 model profile to the llm_minimax node, which aligns with the main objectives across all three modified files.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

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

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[github-actions]

No description provided.

[coderabbitai]

Caution

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

⚠️ Outside diff range comments (1)

tools/src/sync_models.config.json (1)

417-427: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Add protection for MiniMax *-local profiles to avoid Pass-2 deprecation of modelSource: manual entries

tools/src/core/merger.py Pass 2 can deprecate profiles that are missing from the current API response unless they’re included in protected_profile_keys and _source_is_authoritative(deprecation_source, model_source) allows it. For deprecation_source == "provider API", _source_is_authoritative treats modelSource of "manual" (default when unset) as authoritative—so the unprotected MiniMax local profiles (minimax-m2-local, minimax-m2-5-local, minimax-m2-7-local in services.json) are at risk on no-key/fallback sync runs. Add them to protected_profiles or explicitly set a non-deprecating modelSource.
-->

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tools/src/sync_models.config.json` around lines 417 - 427, Add the missing
MiniMax local profiles to the protected_profiles list so they aren't pass-2
deprecated when modelSource defaults to "manual": append entries for
"minimax-m2-local", "minimax-m2-5-local", and "minimax-m2-7-local" to the
protected_profiles array (or alternatively ensure those profiles in the source
have an explicit non-manual modelSource), referencing the protected_profiles key
in the config and the MiniMax profile names to prevent deprecation during
merger.py's Pass 2 logic.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Outside diff comments:
In `@tools/src/sync_models.config.json`:
- Around line 417-427: Add the missing MiniMax local profiles to the
protected_profiles list so they aren't pass-2 deprecated when modelSource
defaults to "manual": append entries for "minimax-m2-local",
"minimax-m2-5-local", and "minimax-m2-7-local" to the protected_profiles array
(or alternatively ensure those profiles in the source have an explicit
non-manual modelSource), referencing the protected_profiles key in the config
and the MiniMax profile names to prevent deprecation during merger.py's Pass 2
logic.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: 96510449-3f6d-4dd7-8b34-5c1b19684de3

📥 Commits

Reviewing files that changed from the base of the PR and between 9a0b78f and 90581fd.

📒 Files selected for processing (3)

• nodes/src/nodes/llm_minimax/README.md
• nodes/src/nodes/llm_minimax/services.json
• tools/src/sync_models.config.json