Back to browse
GitHub Repository

Low-latency IPC library for building persistent agentic tool servers (LLM inference, TTS, vector search, browser automation) over named pipes on the same machine.

15 starsPython

A better alternative to CLI and MCP for local tools

by stefanwebb·Apr 13, 2026·2 points·0 comments

AI Analysis

●●SolidBig BrainNiche Gem

Named pipes beat HTTP latency for local agentic tools while keeping in-memory state between calls.

Strengths
  • Persistent servers hold model weights and indexes in memory instead of reloading per-call.
  • cpipe CLI utility enables ad-hoc commands to any running pipe server from terminal.
  • Claude Code skill teaches agents to discover and query running pipe servers automatically.
Weaknesses
  • Unix-only limits audience; no Windows named pipe equivalent implementation.
  • Agentic tools category is emerging with unclear long-term adoption patterns.
Target Audience

Developers building local AI agents, voice assistants, and LLM tool servers

Similar To

MCP · LangChain Tools

Post Description

I've created an alternative to CLI and MCP for locally running agentic tools. It uses Unix-based OS's named pipes, which means the client has quick IPC with the tool and it can have in-memory state. I created it to build components for voice agents (TTS, STT, LLM chat) but it has wide range of uses! Comes with a CLI tool, cpipe (like curl), and a Claude Code skill

Similar Projects

Developer Tools●●Solid

Xlg – Jq for APIs

Browse Reddit and GitHub via Unix pipes plus MCP server for agents.

Solve My ProblemNiche Gem
alykirk
112mo ago
Developer Tools●●Solid

Prism MCP v4.0 – Behavioral Memory for AI Agents

Persistent memory and time travel for AI agents using local SQLite.

WizardryNiche Gem
dcostenco
102mo ago