Summary

ruvnet/agent-harness-generator provides a meta-tool for creating custom AI agent harnesses. It addresses the need for developers to quickly scaffold branded, repo-aware agent environments with features like custom CLIs, MCP servers, scoped memory, and secure release processes. Targeting engineers and power users, it enables the generation of deployable agent packages that integrate with various LLM providers, offering a factory approach to agent framework creation.

How It Works

This project acts as a factory, generating bespoke AI agent harnesses rather than being a framework itself. Users mint harnesses from GitHub repos or blank slates via an npx CLI or browser Studio. The output is an npm-publishable package with a custom CLI, agents, repo-derived skills, scoped memory, and governance policy. It supports multiple AI hosts and includes Darwin Mode for self-evolving harnesses and @metaharness/router for optimizing model selection. The core advantage is enabling users to own, brand, and customize their agent solutions.

Quick Start & Requirements

• Primary Install/Run:
  • Browser Studio (zero install): https://ruvnet.github.io/agent-harness-generator/
  • CLI Generation: npx metaharness <your-bot-name> --template vertical:coding --host claude-code
  • Interactive Wizard: npx metaharness --wizard
• Prerequisites: Node.js environment for CLI usage.
• Links: Studio: https://ruvnet.github.io/agent-harness-generator/, User Guide: docs/USERGUIDE.md.

Highlighted Details

• Customizable Output: Generates an npm-publishable .zip with user branding and an npx <your-name> CLI, allowing targeted agent solutions by trimming unused components.
• Broad Host Integration: Supports Claude Code, OpenAI Codex, pi.dev, Hermes, OpenClaw, RVM, GitHub Copilot, and GitHub Actions.
• Cost Optimization: Features @metaharness/router to intelligently select the most cost-effective LLM for a task, trained on user data.
• Enhanced Security: Implements witness-signed releases (Ed25519), default-deny MCP security with static scanning (harness mcp-scan), and generates SPDX-2.3 SBOMs.
• Repo Analysis: Pre-scaffolding tools like npx metaharness score <repo> provide suitability reports without executing code.

Maintenance & Community

The project maintains a robust CI matrix across Rust, WASM, and Node.js, with a mature release pipeline. Security is addressed via a dedicated pipeline including cargo-audit, npm-audit, and CodeQL. It is in a v0.1.x beta phase, with ongoing documentation work. Direct community links are not explicitly provided.

Licensing & Compatibility

Licensed under the MIT license, which is permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

The project is in a v0.1.x beta status, indicating potential for ongoing changes. Documentation and credibility reconciliation are noted as active development areas. Analysis tools are static-only, ensuring safety during pre-scaffolding.