CONFIG.md

Configuration

It is recommended to fill in environment variables in the Workers configuration interface, rather than directly modifying variables in the JS code.

KV Configuration

KEY	Special Explanation
DATABASE	Create KV first, name it arbitrarily, and then set it to DATABASE when binding.

System Configuration

Configuration that is common to each user, usually filled in the Workers configuration interface.

KEY	Description	Default Value	Special Description
LANGUAGE	Language	zh-CN	zh-CN，zh-TW, en
AI_PROVIDER	AI provider	auto	AI providers: auto, azure, openai, workers; auto automatically selects a valid configuration. The priority order is azure > openai > workers.
UPDATE_BRANCH	Git branch	master	Branch where version detection is located
-	-	-	-
TELEGRAM_API_DOMAIN	Telegram	https://api.telegram.org	Customization of Telegram API server.
TELEGRAM_AVAILABLE_TOKENS	Support for multiple Telegram Bot Token	null	Multiple Token separated by ,
-	-	-	-
CHAT_WHITE_LIST	Chat ID Whitelisting	null	Multiple IDs are separated by ,, not knowing the IDs, talking to the bot for a sentence returns
I_AM_A_GENEROUS_PERSON	Close the whitelist and allow access to all	false	Since many people don't want to whitelist, or don't know how to get an ID, setting this option will allow everyone to access it, with a value of true.
LOCK_USER_CONFIG_KEYS	Lock custom user configurations	[]	You can lock certain fields. For example, setting it to CHAT_MODEL can prevent other users from switching models through the /setenv command. Multiple fields are separated by ,.
-	-	-	-
AUTO_TRIM_HISTORY	Automatically trim history	true	To avoid the 4096 character limit, truncate the message
MAX_HISTORY_LENGTH	Maximum history length	20	When AUTO_TRIM_HISTORY is turned on To avoid the 4096 character limit, truncate the message
MAX_TOKEN_LENGTH	Maximum number of historical tokens	2048	Too long and easy to timeout suggest setting at a suitable number
GPT3_TOKENS_COUNT	GTP counting mode	false	Use more accurate token counting mode instead of just judging string length, but it's easy to time out
GPT3_TOKENS_COUNT_REPO	GPT3 Counter Resource Repo	https://raw.githubusercontent.com/tbxark-arc/GPT-3-Encoder/master	Resource file for loading GPT3 Token counting configurations
-	-	-	-
SYSTEM_INIT_MESSAGE	System initialization message	You are a useful assistant.	Default robot init message
SYSTEM_INIT_MESSAGE_ROLE	System initialization message Role	system	Default robot init role
-	-	-	-
ENABLE_USAGE_STATISTICS	Enable usage statistics	false	After enabling, each API call will be recorded in KV and can be viewed through /usage.
HIDE_COMMAND_BUTTONS	Hide command buttons	null	Write the buttons you want to hide separated by commas /start,/system, remember to include slashes, and after modifying, reinitialize init.
SHOW_REPLY_BUTTON	Show Quick Reply button	false	Display quick reply buttons.
-	-	-	-
DEBUG_MODE	Debug mode	false	Currently, the latest message can be saved to KV for convenient debugging. It consumes a lot of KV write volume and must be turned off in the production environment.
DEV_MODE	Developer mode	false	Development testing
STREAM_MODE	Stream mode	true	Get a typewriter output mode similar to ChatGPT Web.
SAFE_MODE	Safe mode	true	Safe mode will increase KV write overhead, but it can avoid Telegram's death loop retry caused by Workers timeout, reduce token waste, and it is not recommended to disable.
-	-	-	-
API_KEY	OpenAI API Key	null	Multiple keys can be used at the same time, and one will be randomly selected when using the
CHAT_MODEL	Open AI model	gpt-3.5-turbo
OPENAI_API_DOMAIN	OpenAI API Domain [Deprecated: use OPENAI_API_BASE]	https://api.openai.com	Can be replaced with the domain name of other service providers compatible with OpenAI API.
OPENAI_API_BASE	OPENAI API Base URL	https://api.openai.com/v1	Compatible with Cloudflare AI Gateway
-	-	-	-
AZURE_API_KEY	azure api key	null	Support Azure API, choose either of the two keys. If you want to use Azure by default, you can set AI_PROVIDER to azure.
AZURE_COMPLETIONS_API	azure completions api url	null	https://YOUR_RESOURCE_NAME.openai.azure.com/openai/deployments/YOUR_DEPLOYMENT_NAME/chat/completions?api-version=2023-05-15
AZURE_DALLE_API	azure dalle api url	null	https://YOUR_RESOURCE_NAME.openai.azure.com/openai/deployments/dall-e-3/images/generations?api-version=2023-12-01-preview
-	-	-	-
CLOUDFLARE_ACCOUNT_ID	Cloudflare User ID	null	You can find this information in the right sidebar of the workers homepage, If you want to use Azure by default, you can set AI_PROVIDER to workers.
CLOUDFLARE_TOKEN	Cloudflare Token	null	You can create using the Workers AI (Beta) template at https://dash.cloudflare.com/profile/api-tokens.
WORKERS_CHAT_MODEL	Text generation model	@cf/mistral/mistral-7b-instruct-v0.1	You can check the specific model list at https://developers.cloudflare.com/workers-ai/models/llm/.
WORKERS_IMAGE_MODEL	Text-to-image generation model	@cf/stabilityai/stable-diffusion-xl-base-1.0	Same as above.
-	-	-	-
DALL_E_MODEL	The model of the generated image	dall-e-2	Support dall-e-2 and dall-e-3
DALL_E_IMAGE_SIZE	The size of the generated image	512x512	The size of the generated images. Must be one of 256x256, 512x512, or 1024x1024 for dall-e-2. Must be one of 1024x1024, 1792x1024, or 1024x1792 for dall-e-3 models.
DALL_E_IMAGE_QUALITY	The image quality of the generated image	standard	The quality of the image that will be generated. hd creates images with finer details and greater consistency across the image. This param is only supported for dall-e-3.
DALL_E_IMAGE_STYLE	The image style of the generated image	vivid	The style of the generated images. Must be one of vivid or natural. Vivid causes the model to lean towards generating hyper-real and dramatic images. Natural causes the model to produce more natural, less hyper-real looking images. This param is only supported for dall-e-3.

Group Configuration

You can add the bot to a group, and then everyone in the group can chat with the bot.

BREAKING CHANGE: Major changes, you must add the group ID to the whitelist CHAT_GROUP_WHITE_LIST to use it, otherwise anyone can add your bot to the group and consume your quota.

IMPORTANT: Due to the privacy and security policies of restricted Telegram groups, if your group is a public group or has more than 2000 people, please set the bot as administrator, otherwise the bot will not respond to chat messages with @bot.

IMPORTANT：Must set /setprivacy to disable in botfather, otherwise the bot will not respond to chat messages with @bot.

KEY	Explanation	Default Value	Special Explanation
GROUP_CHAT_BOT_ENABLE	Enable group chat bot	true	After enabling, the bot joins the group and everyone in the group can chat with the bot.
TELEGRAM_BOT_NAME	Bot name xxx_bot	null	The order must be consistent with TELEGRAM_AVAILABLE_TOKENS, must be set, otherwise it cannot be used in group chat
GROUP_CHAT_BOT_SHARE_MODE	Share chat history of group chat bot	false	After enabling, the group has only one session and configuration. If disabled, each person in the group has their own session context.
CHAT_GROUP_WHITE_LIST	Group chat ID whitelist	null	Separate multiple IDs with ,. If you don't know the ID, you can chat with the bot in the group and it will return it.

User Configuration

Each user's custom configuration can only be modified by sending a message through Telegram. The message format is /setenv KEY=VALUE. User configurations have higher priority than system configurations. If you want to delete a configuration, please use /delenv KEY. To set variables in bulk, please use /setenvs {"KEY1": "VALUE1", "KEY2": "VALUE2"}.

KEY	Description
AI_PROVIDER	Configuration same as ENV.AI_PROVIDER
CHAT_MODEL	Configuration same as ENV.CHAT_MODEL
OPENAI_API_KEY	After setting this value, the system-configured KEY will not be called.
OPENAI_API_EXTRA_PARAMS	OpenAI API additional parameters, once set, will be included in every API call and can be used to adjust temperature and other parameters. /setenv OPENAI_API_EXTRA_PARAMS={"temperature": 0.5} Each modification must be a complete JSON.
SYSTEM_INIT_MESSAGE	Configuration same as ENV.SYSTEM_INIT_MESSAGE
DALL_E_MODEL	Configuration same as ENV.DALL_E_MODEL
DALL_E_IMAGE_SIZE	Configuration same as ENV.DALL_E_IMAGE_SIZE
DALL_E_IMAGE_QUALITY	Configuration same as ENV.DALL_E_IMAGE_QUALITY
DALL_E_IMAGE_STYLE	Configuration same as ENV.DALL_E_IMAGE_STYLE
AZURE_API_KEY	Configuration same as ENV.AZURE_API_KEY
AZURE_COMPLETIONS_API	Configuration same as ENV.AZURE_COMPLETIONS_API
AZURE_DALLE_API	Configuration same as ENV.AZURE_DALLE_API
WORKERS_CHAT_MODEL	Configuration same as ENV.WORKERS_CHAT_MODEL
WORKER_IMAGE_MODEL	Configuration same as ENV.WORKER_IMAGE_MODEL
GOOGLE_API_KEY	Configuration same as ENV.GOOGLE_API_KEY
GOOGLE_COMPLETIONS_API	Configuration same as ENV.GOOGLE_COMPLETIONS_API
GOOGLE_COMPLETIONS_MODEL	Configuration same as ENV.GOOGLE_COMPLETIONS_MODEL

Support command

Command	Description	Example
/help	Get command help.	/help
/new	Initiate a new conversation.	/new
/start	Get your ID and start a new conversation.	/start
/img	Generate an image.	/img Image description
/version	Get the current version number and determine if an update is needed.	/version
/setenv	Set user configuration, see User Configuration for details.	/setenv KEY=VALUE
/delenv	Delete user configuration.	/delenv KEY
/usage	Get the usage statistics of the current robot.	/usage
/system	View some current system information.	/system
/role	Set the preset identity, configure usage method same as /setenv.	/role
/redo	Modify the previous question or provide a different answer.	/redo modified content or /redo
/echo	Echo message, only available in development mode.	/echo

Custom Commands

In addition to the instructions defined by the system mentioned above, you can also customize shortcut commands, which can simplify certain longer commands into single-word instructions.

Custom commands use environment variables to set CUSTOM_COMMAND_XXX, where XXX is the command name, such as CUSTOM_COMMAND_azure, and the value is the command content, such as /setenvs {"AI_PROVIDER": "azure"}. This allows you to use /azure instead of /setenvs {"AI_PROVIDER": "azure"} for quick switching of AI providers.

Here are some examples of custom commands.

command	value
CUSTOM_COMMAND_azure	/setenvs {"AI_PROVIDER": "azure"}
CUSTOM_COMMAND_workers	/setenvs {"AI_PROVIDER": "workers"}
CUSTOM_COMMAND_gpt3	/setenvs {"AI_PROVIDER": "openai", "CHAT_MODEL": "gpt-3.5-turbo"}
CUSTOM_COMMAND_gpt4	/setenvs {"AI_PROVIDER": "openai", "CHAT_MODEL": "gpt-4"}

If you are using toml for configuration, you can use the following method:

CUSTOM_COMMAND_azure= '/setenvs {"AI_PROVIDER": "azure"}'
CUSTOM_COMMAND_workers = '/setenvs {"AI_PROVIDER": "workers"}'
CUSTOM_COMMAND_gpt3 = '/setenvs {"AI_PROVIDER": "openai", "CHAT_MODEL": "gpt-3.5-turbo"}'
CUSTOM_COMMAND_gpt4 = '/setenvs {"AI_PROVIDER": "openai", "CHAT_MODEL": "gpt-4"}'