Shim

Shim is a tiny, local-first command-line tool connecting users to various models, MCP servers, and agent skills.

Shim serves as a small adapter between users and model-agnostic coding agents. It provides a scriptable CLI and processes ordinary language commands using a local mini-model router. Users can bring any model API through native OpenAI-compatible and Anthropic adapters or presets like Ollama, LM Studio, and OpenAI. It supports MCP servers over stdio and HTTP, alongside lazy loading of SKILL.md files. The project contains no Electron, Node, Python, Docker, database, or background daemon.

Shim at a glance

Best for: Software engineer
Runs on: macOS, Windows, Linux, CLI
Pricing: Free and open source under the MIT license
Released: November 2026
Category: AI coding tool
Website: github.com/balanov/shim

Key features

Run as a static binary with zero dependencies
Translate management requests via a local mini-model
Connect to OpenAI and Anthropic model APIs
Integrate MCP servers over stdio and HTTP
Load SKILL.md files on demand
Store configuration in plain JSON and sessions in JSONL
Enforce safety permissions requiring confirmation for mutations

Use it for

Manage coding agents from the terminal
Route natural language commands to allowed actions
Load custom agent skills from local directories
Connect disparate model providers and tools

Alternatives to Shim

SmallCode — Runs fully local with budget-managed context and a forgiving parser for small models.
jevgrep — Reduces coding agent costs by providing targeted, relevant repository context.
Little Coder — Optimized for small local models with built-in planning and research tools.
APM — Standardize and reproduce AI agent configurations with dependency management.
LangChain — Compose exactly the agent your use case needs from model, tools, prompt, and middleware.
mini-SWE-agent — It offers high performance with a simplified, minimal codebase.
HarnessRouter — One API to manage agent runtimes, sandboxes, and execution lifecycles.
ZCode — Integrates GLM-5.3 with multi-agent workflows and remote chat control.
Skillhub Local — Consolidates scattered AI agent context and skills into one local dashboard.
Harness Manager — Manages your entire AI coding stack in one native Mac app.
RubyLLM — It provides a unified API for 17 AI providers with native Ruby and Rails integration.
Zero — Operates entirely on the local machine with configurable permission-gated actions.

All Shim alternatives