TextPrompts Documentation
TextPrompts Documentation
A minimal, zero-dependency prompt loader for Python that keeps your prompts close to your code but out of your source files.
Overview
TextPrompts solves the common problem of managing prompts in AI applications. Instead of embedding prompts in your source code (where they get mangled by formatters) or using complex prompt management systems, TextPrompts lets you store prompts in simple .txt files with optional metadata.
Key Features
- Zero runtime dependencies (except Pydantic v2)
- Simple TOML frontmatter with
---delimiters - PromptString class prevents format() errors
- Recursive directory loading with glob patterns
- Performance safeguards with configurable limits
- Comprehensive error handling
- Full type hints and IDE support
Quick Example
prompt.txt:
---
title = "Customer Support"
version = "1.0.0"
---
Hello {customer_name},
Thank you for contacting us about {issue_type}.
We'll resolve this promptly.
Best regards,
{agent_name}
Python code:
from textprompts import load_prompt, set_metadata
# Optionally, configure metadata parsing
set_metadata("allow") # optional!
# Or set via environment variable before import
# export TEXTPROMPTS_METADATA_MODE=allow
prompt = load_prompt("prompt.txt")
response = prompt.prompt.format(
customer_name="Alice",
issue_type="billing",
agent_name="Bob"
)
Installation
pip install textprompts
# or
uv add textprompts