Awesome-LLM-Prompt-Optimization: a curated list of advanced prompt optimization and tuning methods in Large Language Models

Superpipe - optimized LLM pipelines for structured data

Implementation of the paper Fast Inference from Transformers via Speculative Decoding, Leviathan et al. 2023.

Nadir is a Python package designed to dynamically choose the best llm for your prompt by balancing complexity and cost and response time.

Optimized fork of Superpowers — Retains full original features while adding automatic 3-tier workflow routing, integrated safety guards (OWASP-aligned), red-team adversarial testing with auto-fix pipeline, and an estimated 15-30% session overhead reduction (research-informed). Delivers faster, more reliable, hallucination-resistant coding sessions.

DSPEx - Declarative Self-improving Elixir | A BEAM-Native AI Program Optimization Framework

A Demo of Running Sleep-time Compute to Reduce LLM Latency

Declarative Self Improving Elixir - DSPy Orchestration in Elixir

⚡ Cut LLM inference costs 80% with Programmatic Tool Calling. Instead of N tool call round-trips, generate JavaScript to orchestrate tools in Vercel Sandbox. Supports Anthropic, OpenAI, 100+ models via AI Gateway. Novel MCP Bridge for external service integration.

Credit Optimizer v5 for Manus AI - Save 30-75% on credits. Free MCP Server (PyPI) + $9 Manus Skill. Audited across 53 scenarios. Zero quality loss.

Generate clean, AI-ready llms.txt files for your website or docs. Supports crawling, sitemaps, static builds, and framework-aware adapters (Next.js, Vite, Nuxt, Astro, Remix). Includes Markdown/MDX docs mode and robots.txt generator for LLM and search crawlers.

Hybrid adaptive memory system for Claude Code — Cheatsheet (positive patterns) + Immune (negative patterns)

A comprehensive toolkit for training and running lightweight adapters for GGUF-based language models (ERNIE, Llama, Mistral, Phi-3, etc.) without modifying the base model.

Opti-Oignon is a comprehensive optimization framework for local LLMs running on Ollama. It maximizes the performance of your local models through intelligent task routing based on a custom benchmark, RAG (Retrieval-Augmented Generation), and multi-model orchestration.

End-to-End Python implementation of CompactPrompt (Choi et al., 2025): a unified pipeline for LLM prompt and data compression. Features modular compression pipeline with dependency-driven phrase pruning, reversible n-gram encoding, K-means quantization, and embedding-based exemplar selection. Achieves 2-4x token reduction while preserving accuracy.

Adaptive semantic cache for LLMs with streaming support, ML-based thresholds, and real-time cost tracking. Built in Rust for sub-millisecond performance.

Bio-inspired optimization pipeline for Claude Code. 3 systems: Slime Mold (explore→prune) → PRISM (perspectives→compile) → Immune (scan→correct→learn). Domain-agnostic.

Structured repository covering LLM foundations, fine-tuning workflows, optimization strategies, deployment patterns, evaluation methods, and Responsible AI considerations.

LLM context compression proxy — 40-70% token savings, zero code changes

An intelligent prompt optimizer that tailors your prompts for specific LLMs like Gemini, Claude (Anthropic), and ChatGPT using advanced prompt engineering techniques.