lexigram-ai-prompt

AI prompt management for the Lexigram Framework — templates, composition, optimization

Overview

Type-safe prompt template management for the Lexigram AI framework. Build, version, compose, and auto-optimize prompts via DI — with injection protection, multi-format rendering, and a DSPy-inspired optimizer built in. Zero-config usage starts with sensible defaults.

Full documentation: docs.lexigram.dev

Install

uv add lexigram-ai-prompt
# Optional extras
uv add "lexigram-ai-prompt[jinja2]"

Quick Start

from lexigram import Application
from lexigram.di.module import Module, module

from lexigram.ai.prompt import PromptModule
from lexigram.ai.prompt.config import PromptConfig

@module(imports=[
    PromptModule.configure(
        PromptConfig(default_format="f_string", sanitize_inputs=True)
    )
])
class AppModule(Module):
    pass

app = Application(modules=[AppModule])
if __name__ == "__main__":
    app.run()

Configuration

Zero-config usage: Call PromptModule.configure() with no arguments to use defaults.

Option 1 — YAML file

# application.yaml
ai_prompt:
  enabled: true
  default_format: "f_string"
  sanitize_inputs: true
  strict_sanitizer: true
  max_variable_length: 0

Option 2 — Profiles + Environment Variables (recommended)

export LEX_AI_PROMPT__DEFAULT_FORMAT=jinja2
# Environment variables for each field

Option 3 — Python

from lexigram.ai.prompt.config import PromptConfig
from lexigram.ai.prompt import PromptModule

config = PromptConfig(
    default_format="jinja2",
    sanitize_inputs=True,
    strict_sanitizer=True,
    max_variable_length=4096,
)
PromptModule.configure(config)

Config reference

Module Factory Methods

Key Features

• Template types: StringPromptTemplate, ChatPromptTemplate, FewShotPromptTemplate, PartialPromptTemplate
• Rendering formats: f_string, jinja2, dollar, and simple templates
• Registry: Named prompt lookup via PromptRegistry
• Versioning: VersionedPromptStore with history and rollback
• Composition: PromptPipeline (sequential) and ConditionalPrompt (branching)
• Optimizer: DSPy-inspired automatic prompt improvement with BOOTSTRAP_FEW_SHOT, TEMPLATE_REFINEMENT, and ENSEMBLE strategies

Testing

async with Application.boot(modules=[PromptModule.stub()]) as app:
    # your test code
    ...

Key Source Files