ECC

About ECC

ECC (Efficient Claude Code) is an open-source performance optimization framework for AI coding agents. It analyzes agent execution patterns, identifies token waste, reduces redundant tool calls, and optimizes prompt routing to cut costs and improve speed. With 199K+ GitHub stars and +918 stars/day velocity, it has become the standard performance layer for teams running agentic coding at scale. ECC works as a middleware between your agent harness and the underlying model API, providing real-time profiling dashboards, automatic prompt compression, and intelligent caching of repeated tool call patterns. Built by affaan-m, it supports Claude Code, Codex, Gemini CLI, and any OpenAI-compatible endpoint.

Key Features

• Agent execution profiling — identifies token waste and redundant tool calls
• Automatic prompt compression and caching for repeated patterns
• Real-time cost and performance dashboards
• Support for Claude Code, Codex, Gemini CLI, and OpenAI-compatible endpoints
• Middleware architecture — drops into existing harness setups without code changes
• Batch optimization recommendations with before/after token comparisons