t-prompts

Provenance-preserving prompts for LLMs using Python 3.14's template strings

What is t-prompts?

t-prompts turns Python 3.14+ t-strings into navigable trees that preserve full provenance information (expression text, conversions, format specs). Perfect for building, composing, and auditing LLM prompts.

Unlike f-strings which immediately evaluate to strings, t-prompts keeps the structure intact so you can:

• Trace exactly which variable produced which part of your prompt
• Navigate nested prompt components programmatically
• Compose complex prompts from smaller, reusable pieces
• Audit with complete provenance for compliance and debugging
• Validate types at prompt creation (no accidental str(obj) surprises)

Requirements: Python 3.14+

Quick Example

from t_prompts import prompt

# Create a structured prompt
instructions = "Always answer politely."
p = prompt(t"Obey {instructions:inst}")

# Renders like an f-string
print(str(p))  # "Obey Always answer politely."

# But preserves full provenance
node = p['inst']
print(node.expression)  # "instructions" (original variable name)
print(node.value)       # "Always answer politely."

This enables riching tooling:

Targeted Use Cases

• Prompt Debugging: "What exactly did this tangle of code render to?"
• Prompt Optimization (Performance): "What wording / content best achieves my goal?"
• Prompt Optimization (Size): "How do I get the same result with fewer words?"
• Prompt Compacting: "LLM tells me to keep it short, now what do I do?"

Caveats

While this library targets the creation of structured multi-modal prompts, despite the name, there is nothing in particular tying this library to LLMs / Generative models. (It is more "t" than "prompts") To use it for an actual LLM call, you would need to convert the IR into a model specific form (though for text, it could be as simple as str(prompt))

Documentation

📚 Full documentation: https://habemus-papadum.github.io/t-prompts/

• Quick Interactive Demo
• Tutorial

Installation

pip install t-prompts

Or with uv:

uv add t-prompts

Development

This project uses UV and PNPM for dependency management.

# Onetime Setup (per dev machine)
curl -LsSf https://astral.sh/uv/install.sh | sh # Other options exist!
curl -fsSL https://get.pnpm.io/install.sh | sh - # Other options exist!
./scripts/setup-visual-tests.sh

# Repo setup (per clone)
./scripts/setup.sh

# Lint and format
uv run ruff check .
uv run ruff format .

# Build documentation
uv run mkdocs serve

See Developer Setup for detailed instructions.

License

MIT License - see LICENSE file for details.