Untether

Untether is an open-source agent harness and orchestrator developed by Little Bear Apps (littlebearapps). Built primarily in Python, it provides tooling and structure for managing autonomous coding agent workflows and decoupling development tasks across different coding assistant frameworks. In day-to-day operation, Untether allows developers to run and orchestrate agent tasks from the command line. The repository structure indicates direct integration and compatibility with leading frontier models and developer agent configs, including Anthropic Claude (.claude, CLAUDE.md) and OpenAI Codex (.codex), alongside configuration rules for environments like Cursor (.cursorrules). What distinguishes Untether is its focus on serving as a dedicated harness rather than a single monolithic coding assistant. By formalizing agent configurations through dedicated agent definition files (such as AGENTS.md) and harness scripts, it coordinates repetitive engineering tasks and multi-step workflows across repositories. Untether is open source and hosted publicly on GitHub under standard community licensing. As of version 0.35.x, it represents an active project tailored for developers building automated agent pipelines.

Tags: ai agent, cli, coding agent, developer tools, OpenSource

Visit Untether

Key features

  • Multi-Agent Configuration Support — Maintains compatibility profiles and instruction files for major agent formats including Claude (CLAUDE.md) and Cursor (.cursorrules). - Dedicated Agent Harness — Provides an orchestration layer in src/untether designed to run automated workflows without tight coupling to an IDE. - Codex & Frontier Model Adaptability — Includes configuration hooks for Codex and modern LLM-driven development tooling. - Extensible Scripting Interface — Supplies scriptable pipelines in Python for integrating custom hooks, contributor contributions, and automated tasks. - Repo-Level Instruction Ingestion — Uses structured files such as AGENTS.md to define operational constraints and operational context directly in the repository.

Use cases

Use Case: Standardizing Multi-Agent Context Across Teams A development team using multiple agent tools (Claude Code, Cursor) can use Untether's configuration structure to maintain unified repo guidelines in AGENTS.md and CLAUDE.md, ensuring consistent behavior across different coding harnesses.

Use Case: Headless Agent Execution in CI and Automation Engineers can execute Untether scripts headlessly in CI/CD pipelines or local background jobs to process incoming tasks, run test suites against agent-generated patches, and automate code maintenance.

Use Case: Orchestrating Autonomous Coding Iterations A developer working on complex refactoring tasks can use Untether to harness agent sessions, coordinating prompt templates and workspace context without manual copy-pasting between IDE windows.

Who it is for

Untether is designed for software engineers and teams looking to orchestrate and standardize automated AI coding workflows across multiple agent formats. It is less suited for developers seeking an out-of-the-box GUI chat assistant or simple inline code completion plugin.

Install guide

Installation Install Untether from source or via pip (standard Python package installation based on repository structure): bash git clone https://github.com/littlebearapps/untether.git cd untether pip install -e .

First Run Ensure appropriate environment variables and API credentials are configured for the target models you plan to run (e.g., Anthropic or OpenAI API keys): bash export ANTHROPIC_API_KEY="your-api-key" export OPENAI_API_KEY="your-api-key" Verify that your repository contains relevant agent definition files like AGENTS.md or CLAUDE.md to guide the harness.

Example Session Run an Untether orchestration command to initiate a task harness: bash untether run --task "Refactor test suite to pytest and update fixtures"

Part of CLI Coding Agents

Categories: AI Coding Agents, Code Assistants, Developer Tools

Related CLI agents

  • OpenCode — OpenCode is an open-source AI coding agent developed by Anomaly that operates across terminal environments, desktop applications, and IDE extensions.…
  • goose — goose is an open-source, general-purpose AI agent built in Rust, hosted under the vendor-neutral Agentic AI Foundation (AAIF) at the…
  • SWE-agent — SWE-agent is an open-source autonomous software engineering agent harness built and maintained by researchers from Princeton University and Stanford University.…
  • OpenHands — OpenHands (formerly OpenDevin) is an open-source autonomous AI software development agent and orchestration platform. It provides an extensible runtime environment…
  • Plandex — Plandex is an open-source terminal-based AI coding agent designed to handle complex, multi-file engineering tasks in large codebases. Created by…
  • Weaver — Weaver is an open-source agent harness and orchestration framework maintained by sean35mm on GitHub. Built on top of the Bun…
  • AI Tools
  • Categories
  • Industries
  • CLI Coding Agents
  • MCP Servers
  • MCP Categories