Conversation
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 4 potential issues.
Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
| - tools | ||
| isDeprecated: true | ||
| limits: | ||
| context_window: 132096 |
There was a problem hiding this comment.
Incorrect max_tokens value in command-r-plus config
Medium Severity
The newly added max_tokens: 128000 is inconsistent with max_output_tokens: 4000. Every comparable Cohere model (command-r-08-2024, command-r7b-12-2024, command-r7b-arabic-02-2025) sets max_tokens equal to max_output_tokens (i.e. 4000). The value 128000 matches max_input_tokens, suggesting the wrong field's value was used here.
| - tools | ||
| limits: | ||
| context_window: 16000 | ||
| context_window: 8992 |
There was a problem hiding this comment.
Incorrect context window value for command-a-translate
Medium Severity
The context_window is set to 8992, but the actual context window for command-a-translate-08-2025 is 16K tokens (16384). The correct total context length is 8K input + 8K output = 16K. The value 8992 does not correspond to any documented limit and is almost certainly wrong.
| - tools | ||
| limits: | ||
| context_window: 256000 | ||
| context_window: 288000 |
There was a problem hiding this comment.
Incorrect context window for command-a-03-2025
Medium Severity
The context_window is set to 288000, but the actual context window for command-a-03-2025 is 256K tokens (256000) per Cohere's official documentation. The value 288000 overstates the actual limit and does not match the max_input_tokens: 256000 field in the same file.
| - tools | ||
| limits: | ||
| context_window: 256000 | ||
| context_window: 288768 |
There was a problem hiding this comment.
Incorrect context window for command-a-reasoning
Medium Severity
The context_window is set to 288768, but the actual context window for command-a-reasoning-08-2025 is 256K tokens (256000) per Cohere's official documentation. This is inconsistent with max_input_tokens: 256000 in the same file.
2 participants
Auto-generated by model-addition-agent for provider
cohere.Note
Medium Risk
Primarily updates Cohere model configuration YAMLs, but several models switch
modebetweencompletionandchat, which can change which API/path the runtime uses. Incorrect mode/limit metadata could cause request failures or truncated outputs at runtime.Overview
Updates multiple Cohere model definition YAMLs to align metadata with current capabilities.
Several models adjust their primary
mode(notably switching betweencompletionandchat) and addsupportedModeswhere both are intended. The PR also refreshes declaredfeatures(e.g.,tools/tool_choice/structured_output) and reviseslimits.context_windowfor various chat, embedding, and rerank models.Written by Cursor Bugbot for commit ad58b21. This will update automatically on new commits. Configure here.