fix(kimi-code): correct maxTokens defaults and add server-side override

[myk1yt]

Summary

Fixes #1215

Kimi Code OAuth provider displayed incorrect Max Output values (32,768) for all models. According to Kimi's official documentation, the correct default for kimi-k3 is 131,072.

Root Cause

Three issues in the codebase:

1. kimiCodeDefaultModelInfo.maxTokens was hardcoded to 32_768 instead of 131_072
2. The Zod schema for /models API response didn't include max_tokens field (silently stripped)
3. mapKimiCodeModel() overrode contextWindow from server but not maxTokens

Changes (3 files, +47/-1)

File	Change
packages/types/src/providers/kimi-code.ts	Update maxTokens to 131_072, add kimiCodeModelDefaults for per-model fallback
src/api/providers/fetchers/kimi-code.ts	Add max_tokens to Zod schema, override in mapKimiCodeModel()
src/api/providers/fetchers/__tests__/kimi-code.spec.ts	Add 4 tests for maxTokens override and fallback

Fallback Chain

maxTokens = server.max_tokens
         ?? kimiCodeModelDefaults[model.id].maxTokens
         ?? kimiCodeDefaultModelInfo.maxTokens (131,072)

This ensures automatic updates when the server provides max_tokens, with per-model and global fallbacks.

Verification

• Vitest: 11/11 passed (7 existing + 4 new)
• ESLint: clean
• TypeScript: clean

Summary by CodeRabbit

• New Features

  • Increased the maximum token capacity for supported Kimi Code models to 131,072 tokens.
  • Added model-specific token limits for improved handling of available models.
  • Added support for reading token limits provided by the Kimi Code service.

• Bug Fixes

  • Kimi Code now prioritizes server-provided limits, followed by model-specific and global defaults.
  • Invalid fractional token limits are now rejected.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 291347cc-2bd4-4e44-bd6d-12e02b3a4960

📥 Commits

Reviewing files that changed from the base of the PR and between 568b7c9 and 4cba8fd.

📒 Files selected for processing (2)

• src/api/providers/fetchers/__tests__/kimi-code.spec.ts
• src/api/providers/fetchers/kimi-code.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• src/api/providers/fetchers/kimi-code.ts

---

📝 Walkthrough

Walkthrough

Kimi Code now uses 131,072-token defaults, parses optional server-provided max_tokens, and applies precedence between server values, per-model defaults, and the global fallback. Tests cover each resolution path and reject fractional values.

Changes

Kimi Code max token handling

Layer / File(s)	Summary
Max token defaults and mapping packages/types/src/providers/kimi-code.ts, src/api/providers/fetchers/kimi-code.ts, src/api/providers/fetchers/__tests__/kimi-code.spec.ts	The global Kimi Code default is 131,072 tokens. Four model IDs have per-model defaults. The response schema accepts positive integer max_tokens values. Mapping prioritizes server values, then model defaults, then the global default. Tests cover server overrides, fallback paths, and fractional-value rejection.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• Zoo-Code-Org/Zoo-Code#933: Both changes update Kimi provider model configuration and token-limit handling.
• Zoo-Code-Org/Zoo-Code#945: Both changes update Kimi Code defaults and model discovery mapping.

Suggested labels: awaiting-review

Suggested reviewers: edelauna, taltas

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Linked Issues check	⚠️ Warning	The PR fixes schema parsing, server overrides, and global defaults, but all four model fallbacks use 131,072 instead of the documented kimi-for-coding limit.	Set the kimi-for-coding fallback to its documented model-specific limit, or implement an equivalent prompt-aware calculation.
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the Kimi Code maxTokens fix and server-side override.
Description check	✅ Passed	The description explains the issue, implementation, fallback order, affected files, and verification results, although several template sections are omitted.
Out of Scope Changes check	✅ Passed	All production and test changes directly support the linked Kimi Code maxTokens issue and its validation requirements.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

Warning

There were issues while running some tools. Please review the errors and either fix the tool's configuration or disable the tool if it's a critical failure.

🔧 ESLint

If the error stems from missing dependencies, add them to the package.json file. For unrecoverable errors (e.g., due to private dependencies), disable the tool in the CodeRabbit configuration.

src/api/providers/fetchers/__tests__/kimi-code.spec.ts

ESLint skipped: missing config or dependency (missing-dependency). The ESLint configuration references a package that is not available in the sandbox.

src/api/providers/fetchers/kimi-code.ts

ESLint skipped: the ESLint configuration for this file references a package that is not available in the sandbox.

---

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.}

[coderabbitai]

Actionable comments posted: 2

🤖 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.

Inline comments:
In `@src/api/providers/fetchers/__tests__/kimi-code.spec.ts`:
- Around line 1-2: Update the fallback assertions in the kimi-code tests around
the cases at lines 124-134 to use independent fixture values for the documented
model defaults, rather than kimiCodeModelDefaults or
kimiCodeDefaultModelInfo.maxTokens. Remove those production-constant imports if
they are no longer used, while preserving the existing fallback behavior checks.

In `@src/api/providers/fetchers/kimi-code.ts`:
- Line 15: Update the max_tokens schema in the Kimi Code response validation to
require positive integers by adding integer validation before the existing
positivity and optional checks. Add a rejection test covering a fractional
max_tokens value and ensure mapKimiCodeModel does not receive it as
ModelInfo.maxTokens.

🪄 Autofix

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: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 072dd6a8-85c9-4888-bf7b-29f3278580fb

📥 Commits

Reviewing files that changed from the base of the PR and between fcc507c and 568b7c9.

📒 Files selected for processing (3)

• packages/types/src/providers/kimi-code.ts
• src/api/providers/fetchers/__tests__/kimi-code.spec.ts
• src/api/providers/fetchers/kimi-code.ts

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[myk1yt]

@taltas I am so sorry! I didn't know how to handle the changes in my fork, so I made a mistake. Could you please check/help again?