Prompt Optimization Python Packages

agentdescent

Gradient descent, but the parameters are agents — a parallel, asynchronous framework for self-evolving agents (skills, prompts, harnesses). Diffs are the gradients; the aggregator is the optimizer.

textgrad

TextGrad: Automatic ''Differentiation'' via Text -- using large language models to backpropagate textual gradients. Published in Nature.

prompt-flamegraph

Lightweight, zero-dependency Python package to profile LLM prompt tokens with interactive flamegraphs, waste detection, prompt diffs and HTML/SVG/Markdown/terminal exports.

weco

Production-Grade Autoresearch. Ideal for agent harness engineering, prompt engineering, ML model development, GPU kernels, and other optimizable code.

jev-align

Build calibrated AI Functions from human feedback using Jev and GEPA.

evalfloor

EvalFloor: is your LLM eval improvement real? Trying k prompt variants and keeping the best scores points on noise alone — this computes how many.

functai

Typed functions whose body a language model writes: run them on tables, measure how often they're right, make them better. Python, TypeScript, R and Julia.

shrewd

Turn repeated LLM classification calls into a small, fast, local model for one fixed task.

nativeprompt

Переписывает промпт под правила модели, где вы работаете (Claude Opus 5.5, Fable 5.1, GPT-6 Astra/Sol/Luna, Codex), и подсказывает, на каком уровне усилий её запускать. Каждый совет со ссылкой на официальную документацию. Без API-ключей, без зависимостей, офлайн.

evalt

Budget-bounded LLM routing that finds the cheapest model and prompt meeting your accuracy target.

promptuna-server

An evaluation and optimization harness for functions that use Language Models.

promptuna-cli

An evaluation and optimization harness for functions that use Language Models.

synth-optimizers

GEPA and GELO optimizer runbooks, SDKs, and hosted optimizer surfaces for Synth.

omegaprompt

The overfit gate for your prompts: re-test the winning prompt on held-out examples it never tuned on, and block the ship if it doesn't generalize. Sits on top of promptfoo/DSPy. CLI + MCP + CI.

axiolex

Shared tool discovery and execution for AI clients and applications across MCP, A2A, and internal tools—intent-ranked Top-K retrieval with centrally managed provider and tool changes

logillm

LogiLLM - Zero-dependency enterprise alternative to dspy

instructvault

Git-native prompt linting, testing, versioning, and CI—without a hosted prompt registry.

trace-opt

End-to-end Generative Optimization for AI Agents

dspy-code

Claude Code for DSPy: Comprehensive CLI to Optimize Your DSPy Code. our AI-Powered DSPy Development Assistant

kochab

A minimal reference implementation of the agent-to-data loop, turning agent evaluations, real-world traces, and human feedback back into training data.

promptcraft

PromptCraft is a prompt perturbation toolkit from the character, word, and sentence levels for prompt robustness analysis. PyPI Package: pypi.org/project/promptcraft

llm-prompt-refiner

🚀 Lightweight Python library for building production LLM applications with smart context management and automatic token optimization. Save 10-20% on API costs while fitting RAG docs, chat history, and prompts into your token budget.

ace-playbook

ACE — Agentic Context Engineering: evolving, self-improving context playbooks for LLM agents. Faithful ICLR 2026 implementation with OpenAI Agents SDK support.

tuneprompt

PromptForge scores LLM prompts across seven quality dimensions, then rewrites weak prompts into clear, intent-preserving instructions