Outlines

Outlines enforces structure at the token level, so your LLM outputs valid JSON, XML, or regex-matched text every time. It works by converting schemas (Pydantic models, JSON schemas, regexes) into finite state machines that filter invalid tokens during generation. You get type-safe outputs with zero parsing errors, and it actually speeds up inference by fast-forwarding through deterministic paths. Works with local models via Transformers, vLLM, or llama.cpp, which matters if you're running 8B models locally and need guaranteed structure without the back-and-forth of prompt engineering. From dottxt.ai with 8,000+ stars. Use it when you're tired of LLMs hallucinating malformed JSON or need production-grade structured extraction.

skills.sh installs: 1K
GitHub stars: 14.4K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add orchestra-research/ai-research-skills --skill outlines --agent claude-code

Installs into .claude/skills of the current project.

Files

Outlines: Structured Text Generation

When to Use This Skill

Use Outlines when you need to:

Guarantee valid JSON/XML/code structure during generation
Use Pydantic models for type-safe outputs
Support local models (Transformers, llama.cpp, vLLM)
Maximize inference speed with zero-overhead structured generation
Generate against JSON schemas automatically
Control token sampling at the grammar level

GitHub Stars: 8,000+ | From: dottxt.ai (formerly .txt)

Installation

# Base installation
pip install outlines

# With specific backends
pip install outlines transformers  # Hugging Face models
pip install outlines llama-cpp-python  # llama.cpp
pip install outlines vllm  # vLLM for high-throughput

Quick Start

Basic Example: Classification

import outlines
from typing import Literal

# Load model
model = outlines.models.transformers("microsoft/Phi-3-mini-4k-instruct")

# Generate with type constraint
prompt = "Sentiment of 'This product is amazing!': "
generator = outlines.generate.choice(model, ["positive", "negative", "neutral"])
sentiment = generator(prompt)

SKILL.md on GitHub