Prompt Partials
With Prompt Partials, you can save your commonly used templates (which could be your instruction set, data structure explanation, examples etc.) separately from your prompts and flexibly incorporate them wherever required.
This feature is available on all Portkey plans.
Partials can also serve as a global variable store. You can define common variables that are used across multiple of your prompt template and can reference or update them easily.
Creating Partials
Partials are directly accessible from the Prompts Page:
You can create a new Partial and use it for any purpose in any of your prompt templates. For example, here’s a prompt partial where we are separately storing the instructions:
Upon saving, each Partial generates a unique ID that you can use inside prompt templates.
Template Engine
Partials also follow the Mustache template engine and let you easily handle data input at runtime by using tags.
Portkey supports {{variable}}
, {{#block}} <string> {{/block}}
, {{^block}}
and other tags.
Check out this comprehensive guide on how to use tags.
Versioning
Portkey follow the same Update
& Publish
flow as prompt templates. You can keep updating the partial and save new versions, and choose to send any version to prod using the Publish
feature.
All the version history for any partial is avaiable on the right column and any previous version can be restored to be latest
or published
to prod easily.
Using Partials
You can call Partials by their ID inside any prompt template by just starting to type {{>
Portkey lists all of the available prompt partials with their names to help you easily pick.
When a partial is incorporated in a template, all the variables/blocks defined are also rendered on the Prompt variables section:
When a new Partial version is Published, your partial that is in use in any of the prompt templates also gets automatically updated.
Making a Prompt Completion Request
All the variables/tags defined inside the partial can now be directly called at the time of making a prompts.completion
request:
const response = portkey.prompts.completions.create({
promptID: "pp-system-pro-34a60b",
variables: {
"user_query":"",
"company":"",
"product":"",
"benefits":"",
"phone number":"",
"name":"",
"device":"",
"query":""
}
})
Was this page helpful?