kb-kimi

Document Kimi K2 Thinking model quirks and workarounds for reliable integration.

97|2|Updated Jan 1, 2026

Search Tags:#debugging #reasoning #bedrock #tool-calls #retry-strategy #kimi-k2 #moonshot-ai

Ready to run

npx skills add https://github.com/minorun365/my-claude-code-settings --skill kb-kimi

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: kb-kimi
Source: https://github.com/minorun365/my-claude-code-settings/tree/main/claude/skills/kb-kimi
Command: npx skills add https://github.com/minorun365/my-claude-code-settings --skill kb-kimi

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Kimi K2 Thinking exhibits model-specific quirks and unpredictable behaviors that differ from Claude, making reliable integration challenging. This Skill documents the differences and practical workarounds to make Kimi K2 flows more stable.

Core Features & Use Cases

Document differences between Claude and Kimi K2 Thinking, including cross-region inference, cache behavior, tool call stability, and think-process exposure.
Provide concrete troubleshooting and retry strategies for unreliable tool calls, reasoning events, and markdown extraction.
Serve as a reference for integrating Kimi K2 thinking into agent workflows with robust guardrails.

Quick Start

Apply the Kimi K2 workarounds in your agent flow whenever you use the moonshot.kimi-k2-thinking model.

Frequently Asked Questions about kb-kimi

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I fix Kimi K2 Thinking tool call failures and reasoning instability?▼

To fix Kimi K2 Thinking tool call failures, apply explicit handling for tool-call reasoning events and implement targeted retry strategies. This documents model quirks and provides workarounds to stabilize unreliable tool calls during agent workflows.

Why does moonshot.kimi-k2-thinking throw AccessDenied errors during inference?▼

Moonshot.kimi-k2-thinking throws AccessDenied errors when certain cache options are enabled. You must avoid these specific cache configurations during cross-region inference to prevent AccessDenied errors and ensure reliable model execution.

What are the differences between Claude and Kimi K2 Thinking for agent workflows?▼

Differences between Claude and Kimi K2 Thinking include distinct cross-region inference, cache behavior, and tool call stability. Kimi K2 exposes think-process events requiring explicit think-tag cleanup and markdown extraction, unlike Claude environments.

How do I clean up think tags and extract markdown from Kimi K2 reasoning events?▼

To clean up think tags and extract markdown from Kimi K2 reasoning events, apply explicit handling logic within your agent flow. This addresses model-specific reasoning exposure quirks to ensure clean markdown extraction.

Can I use cache options with moonshot.kimi-k2-thinking for cross-region inference?▼

You cannot use cache options with moonshot.kimi-k2-thinking for cross-region inference. Applying these cache configurations causes AccessDenied errors, so you must avoid them to maintain stable cross-region operations.