Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/barryceelen/claudette/llms.txt

Use this file to discover all available pages before exploring further.

All Claudette settings live in Claudette.sublime-settings. To open the file, go to Preferences > Package Settings > Claudette > Settings. Settings you define in your user file override the defaults from the package defaults file.

API settings

Type: string or object
Default: ""
Your Anthropic API key. Accepts a plain string for a single key, or an object with a keys array and active_key index for multiple keys.
"api_key": "sk-ant-..."
"api_key": {
  "keys": [
    { "name": "Work", "key": "sk-ant-..." },
    { "name": "Personal", "key": "sk-ant-..." }
  ],
  "active_key": 0
}
See Configure API Keys for details.
Type: string
Default: "https://api.anthropic.com/v1/"
The base URL for all API requests. Change this to use a proxy or a compatible API endpoint.
"base_url": "https://your-proxy.example.com/v1/"
Type: boolean
Default: true
Whether to verify SSL certificates when making API requests. Set to false to allow self-signed certificates on custom base_url endpoints.
"verify_ssl": false
Disabling SSL verification reduces security. Only use this in trusted development environments.
Type: object
Default: {}
Optional custom HTTP headers included in every API request. Useful for proxies that require authentication headers.
"custom_headers": {
  "X-Custom-Header": "value"
}

Model settings

Type: string
Default: "claude-sonnet-4-5"
The Claude model to use for all requests. You can also switch models at runtime with Claudette: Switch Model.
"model": "claude-opus-4-5"
See Choosing Claude Models for a full breakdown and output limits.
Type: integer
Default: 8192
Maximum number of output tokens per response. Must not exceed the selected model’s output limit — the API returns a 400 error if it does.
"max_tokens": 8192
Type: string or number
Default: "1.0"
Range: 0.01.0
Controls response randomness. Lower values produce more deterministic output; higher values produce more varied responses. Values outside the valid range fall back to 1.0.
"temperature": "0.7"

System prompts

Type: array of strings
Default: Three built-in prompts (programming, writing, general)
The list of available system prompts. The active prompt is selected by default_system_message_index.
"system_messages": [
  "You are a helpful AI assistant focused on programming help.",
  "You are a helpful AI assistant focused on writing and documentation.",
  "You are a helpful AI assistant ready to help with any task."
]
Type: integer
Default: 0
Zero-based index of the active system prompt in system_messages. If the index is out of range or the array is empty, no system prompt is sent.
"default_system_message_index": 0

Tools

Type: integer
Default: 5
Range: 120
Maximum number of web searches Claude can make per request when web_search is enabled.
"web_search_max_uses": 5
Type: array of strings
Default: []
Restrict web search results to specific domains. Leave empty to allow all domains. Do not set both web_search_allowed_domains and web_search_blocked_domains at the same time.
"web_search_allowed_domains": ["docs.example.com", "example.com"]
Type: array of strings
Default: []
Exclude specific domains from web search results. Do not set both web_search_blocked_domains and web_search_allowed_domains at the same time.
"web_search_blocked_domains": ["example.com"]
Type: boolean
Default: false
When true, Claude can view and edit files in your project using the text editor tool (view, str_replace, create, insert commands). Uses non-streaming requests. Paths are resolved against project folders or text_editor_tool_roots.
"text_editor_tool": true
Type: integer
Default: not set (no limit)
Maximum number of characters Claude can read when viewing a file (Claude 4 tool only). Set to 0 for no limit. This setting is optional and commented out by default.
"text_editor_tool_max_characters": 10000
Type: array of strings
Default: []
Additional directory paths allowed for file operations. Paths must be under these roots or the current project folders.
"text_editor_tool_roots": ["/home/user/shared-libs"]

Pricing

Type: object
Default: April 2026 rates
Token pricing per 1M tokens, used to calculate and display cost information in the chat view. Contains haiku, opus, and sonnet sub-objects, each with input, output, cache_write, and cache_read fields (USD).
"pricing": {
  "haiku": {
    "input": 1,
    "output": 5,
    "cache_write": 1.25,
    "cache_read": 0.1
  },
  "opus": {
    "input": 5,
    "output": 25,
    "cache_write": 6.25,
    "cache_read": 0.5
  },
  "sonnet": {
    "input": 3,
    "output": 15,
    "cache_write": 3.75,
    "cache_read": 0.3
  }
}
Update these values if Anthropic changes its pricing. Current rates are listed at anthropic.com/pricing.

Chat view

Type: boolean
Default: false
Show line numbers in the chat view.
"chat": {
  "line_numbers": true
}
Type: boolean
Default: false
Show rulers in the chat view.
"chat": {
  "rulers": true
}
Type: boolean
Default: true
When true, the chat view is marked as a scratch buffer and closes without prompting to save.
"chat": {
  "set_scratch": false
}
Type: boolean
Default: true
Show estimated cost information after each response in the chat view.
"chat": {
  "show_cost": false
}

Build docs developers (and LLMs) love