KImi K 2.7 back to back repeated tool calling failures

[aditya-oss-prog]

Description

HI so i am using Kimi K2.7 via together AI and fireworks but it is having back to back massive tool calling failures.

Plugins

No response

OpenCode version

1.17.11

Steps to reproduce

1. Select KImi k2.7 model in huggingface provider list

2. start multiple agents to explore codebase and each agent will have repeated tool calling failures

Screenshot and/or share link

✱Glob ""

• Thought: 169ms
  ✱Glob ""
  Model tried to call unavailable tool 'invalid'. Available tools: bash, glob, grep, read, webfetch.
• Thought: 1.1s
  ✱Glob ""
  The glob tool was called with invalid arguments: SchemaError(Missing key
  at ["pattern"]).
  Please rewrite the input so it satisfies the expected schema.

• Thought: 1.6s
  Use bash to list tests dirs instead of glob? The bash instructions say avoid file ops but ls is okay? They said DO NOT use find/cat but ls is okay for directory verification. I can use ls to list tests directories.
  $
  Model tried to call unavailable tool 'invalid'. Available tools: bash, glob, grep, read, webfetch.

• Thought: 790ms
  $
  Model tried to call unavailable tool 'invalid'. Available tools: bash, glob, grep, read, webfetch.
• Thought: 857ms
  $
  Model tried to call unavailable tool 'invalid'. Available tools: bash, glob, grep, read, webfetch.
• Thought: 3.7s
  $
  Model tried to call unavailable tool 'invalid'. Available tools: bash, glob, grep, read, webfetch.
• Thought: 1.1s
  $
  Model tried to call unavailable tool 'invalid'. Available tools: bash, glob, grep, read, webfetch.
• Thought: 3.5s
  $
  The bash tool was called with invalid arguments: SchemaError(Missing key
  at ["command"]).
  Please rewrite the input so it satisfies the expected schema.
• Thought: 321ms
  $
  The bash tool was called with invalid arguments: SchemaError(Missing key
  at ["command"]).
  Please rewrite the input so it satisfies the expected schema.

Operating System

macos 26.5.1

Terminal

Macos terminal

[github-actions]

This issue might be a duplicate of existing issues. Please check:

• Bug Report: invalid internal tool is exposed and can spam repeated calls #21900: The exact error "Model tried to call unavailable tool 'invalid'" matches this bug report about the invalid internal tool being exposed and causing repeated call loops
• Kimi k2.5 has issues with tool calling #20650: Kimi k2.5 has issues with tool calling — same model family, same tool-calling failure symptoms

If your issue is distinct from these (e.g., specific to K2.7 or a new regression), please add a note explaining what differs.