Fix Cerebras conservative max tokens calculation

[sebastiand-cerebras]

Summary

This PR fixes the conservative max tokens calculation for Cerebras models.

Changes

• Updated the max tokens calculation to use a more conservative approach for Cerebras provider
• Ensures better compatibility and prevents potential issues with token limits

Testing

• Built and verified the extension compiles successfully
• Type checking passes

[changeset-bot]

🦋 Changeset detected

Latest commit: 6fdd55c

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
kilo-code	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[kiloconnect]

✅ No Issues Found

2 files reviewed | Confidence: 95% | Recommendation: Merge

Review Details

Files: src/api/providers/cerebras.ts, .changeset/cerebras-conservative-max-tokens.md

Checked: Security, bugs, performance, error handling

Summary:

• The conservative max_tokens approach (8K default) is well-justified given Cerebras's upfront quota reservation behavior
• The Math.min(max_tokens || CEREBRAS_DEFAULT_MAX_TOKENS, CEREBRAS_DEFAULT_MAX_TOKENS) logic correctly caps tokens at 8K regardless of model's maxTokens setting
• Integration header (X-Cerebras-3rd-Party-Integration: kilocode) added consistently to both createMessage and completePrompt methods
• Constants are well-documented with clear rationale
• No security concerns with the header addition