PromptStore

A lightweight Python package for managing and versioning LLM prompt templates.

Features

• Simple JSON-based storage
• Template versioning
• Tag-based organization
• Jinja2 template syntax
• Variable highlighting in Markdown exports for better visibility
• Package integration utilities

Installation

pip install promptstore

Usage

from promptstore import PromptStore

# Create a store
store = PromptStore("./prompts")

# Add a prompt template
prompt = store.add(
    name="example",
    content="Write a {{language}} function that {{task}}",
    namespace="project1",
    description="Code generation prompt",
    tags=["coding", "generation"],
    subset="subproject1"
)

# Use the prompt
filled = prompt.fill({
    "language": "Python",
    "task": "sorts a list in reverse order"
})

This automatically generates a markdown representation like this:

---
uuid: 77a6b015-efea-436e-8636-6ae06c438ad8
name: example
namespace: project1
description: Code generation prompt
version: 1
tags:
- coding
- generation
variables:
- language
- task
subset: subproject1
created_at: '2025-11-12T11:17:24.128334+00:00'
updated_at: '2025-11-12T11:17:24.128334+00:00'
---

Write a **`{{language}}`** function that **`{{task}}`**

This makes it easy to identify which parts of the prompt are variables when viewing the markdown files.

Prompt Usage

# Fill a prompt template
prompt = store.get("namespace/name@subset")
result = prompt.fill({
    "language": "Python",
    "task": "sorts a list in ascending order"
})

Documentation

Full documentation is available at lamalab-org.github.io/promptstore.

License

This project is licensed under the MIT License - see the LICENSE file for details.