Reference / Prompt Config 52d
Prompt Config
versionstringOptional
Config file version, currently "1".
promptsarrayRequired
A list of prompt definitions.
Prompt object
namestringRequired
Prompt name that will be used as the configuration key in API calls.
descriptionstringRequired
Prompt description that will be shown in the web playground.
providersarrayRequired
A list of prompt parameters for providers.
Provider object
labelstringRequired
Provider label that will be shown in the web playground.
genparamsobjectRequired
LLM call parameters.
Genparam object
Contains LLM provider API parameters.
Refer to the LLM provider API documentation for details, for example:
OpenAI Chat Completion: https://platform.openai.com/docs/api-reference/chat/create
Gemini Text Generation: https://ai.google.dev/gemini-api/docs/text-generation
Anthropic Messages: https://docs.claude.com/en/api/messages
queryparamsarrayoptional
A list of variables to be replaced in the prompt.
Queryparam object
typestringRequired
Variable type.
string: use this for most of the cases
image: for OpenAI Images Edits and Variations API, OpenAI Chat and Gemini image input, as image understanding, the queryparam name is used in prompt config to state the field position, its value is converted to an image object according to LLM SDK.
url: get the URL readable content, not available in DLLM playground, only supported in standalone deployment
htmlsource: get the URL HTML source, not available in DLLM playground, only supported in standalone deployment
namestringRequired
Name of the variable.
creatednumberRequired
Timestamp when the prompt was created.
specconvertbooleanoptional
Reserved.
Parameter Substitution
The config file provides a way to substitute parameters in the prompt.
The variable is defined in brackets, { }, as follows:
Then define the variable types and names in the queryparams array:
Parameter names are passed in the API request body and must not conflict with LLM provider API parameters (e.g., model, messages, prompt). Use descriptive names with underscores for clarity.
Image Input
The queryparams type value image is for image understanding, image edit, image variation. DLLM API supports OpenAI and Gemini image input.
OpenAI prompt example:
Gemini prompt example:
The image input is converted to an image object according to the LLM SDK.