# Kodelyth ECC

> Kodelyth ECC is an open-source MCP server for developer tools. Local toolkit of 71 AI coding agents and an MCP server that cuts token use in coding. It is available on CLI, vscode, macOS, Linux, Windows and self-hosted.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, vscode, macOS, Linux, Windows, self-hosted
- **Website:** https://ecc.kodelyth.com
- **Categories:** [developer tools](https://ai-powered.com/categories/developer-tools)
- **Listing updated:** 2026-11-18

## About

The README and site describe Kodelyth ECC as a layered set of 71 specialist coding agents, 196 skills and 104 slash commands, with a Super Agent that splits a large objective into dependent steps and runs independent ones in parallel. It bundles RTK, which filters shell command output before the model reads it, and Terse, which trims the prose the model writes back, plus an AST codebase graph for structural questions about a repo.

A universal MCP adapter exposes the agents, skills and a local BM25 memory store to clients such as Claude Code, Cursor, Windsurf, Codex and Kiro, and the installer wires them up automatically. Memory Core keys saved solutions to the git remote so recall survives a repo move, and it redacts secrets before writing.

The site says everything runs locally with no telemetry, and the project is MIT licensed and installed through npm. It suits developers who use several AI coding IDEs and want lower token costs. The savings figures are the vendor's own estimates on its sample ledger.

## Features

- **Intent-based agent routing:** Routes a plain-language problem, including pasted stack traces, to one of 71 specialist agents and announces which one took over.
- **Parallel objective orchestration:** Decomposes a large objective into a dependency graph, runs independent steps on separate branches and resumes after a crash.
- **Shell output compression:** RTK runs as a PreToolUse hook to filter and deduplicate command output before it reaches the model.
- **Terse output modes:** Four compression levels reduce the model's written prose while keeping code blocks, commands and paths byte-exact.
- **Codebase knowledge graph:** Builds a tree-sitter AST graph across 158 languages to answer caller and impact questions without file-by-file grep.
- **Red-team security agents:** Eight agents run in parallel to look for prompt injection, supply-chain risks, leaked secrets and backdoors.
- **Local memory recall:** Captures solutions at session end, redacts secrets and recalls relevant past fixes keyed to the git remote.

## Good for

- Cut token costs on daily coding sessions across Claude Code, Cursor or Windsurf
- Break a large refactor into parallel branches with verification steps
- Ask which code paths a change would affect without reading many files
- Run a pre-release security sweep for secrets and supply-chain issues
- Reuse fixes from earlier sessions when working in the same repository

## FAQ

### Is Kodelyth ECC free?

The site says it is MIT licensed, free, requires no account or subscription, and installs with npm i -g kodelyth-ecc.

### Does my code leave my machine?

The site states everything runs locally, including agents, memory, RTK compression and the codebase graph, with zero telemetry and no cloud calls.

### Which IDEs are supported?

The site lists installs for 12 AI coding platforms, including Claude Code, Cursor, Windsurf, Codex CLI, Gemini CLI, Cline, Aider and Kiro.

### How much token saving should I expect?

The site estimates 55-65% total token reduction on coding sessions, based on its own measurements, and says actual savings depend on the workload.

### What are the best alternatives to Kodelyth ECC?

The closest alternatives to Kodelyth ECC are Total Agent Memory, Cork, Memoria — Deterministic AI Memory & Code Intelligence MCP, Chip MCP and keelyard.

## Where this comes from

- Description: written from the project's website.

## Alternatives

- [Total Agent Memory](https://ai-powered.com/mcp/total-agent-memory): Total Agent Memory is an open-source MCP server for knowledge and memory.
- [Cork](https://ai-powered.com/mcp/cork-2): Cork is a freemium MCP server for developer tools.
- [Memoria — Deterministic AI Memory & Code Intelligence MCP](https://ai-powered.com/mcp/memoria-deterministic-ai-memory-and-code-intelligence-mcp): Memoria — Deterministic AI Memory & Code Intelligence MCP is a freemium MCP server for knowledge and memory.
- [Chip MCP](https://ai-powered.com/mcp/chip-mcp): Chip MCP is a freemium MCP server for knowledge and memory.
- [keelyard](https://ai-powered.com/mcp/keelyard): keelyard is an open-source MCP server for developer tools.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [Awareness Local](https://ai-powered.com/mcp/awareness-local): Awareness Local is a freemium MCP server for knowledge and memory.
- [reevesagents](https://ai-powered.com/mcp/reevesagents): reevesagents is an open-source MCP server for developer tools.

---
Source: https://ai-powered.com/mcp/kodelyth-ecc
