Summary

The strongdm/attractor repository provides Natural Language Specifications (NLSpecs) for building "Attractor," a non-interactive coding agent designed for a "Software Factory" paradigm. This approach automates software development by having agents write and validate code based on human-readable specs and scenarios, eliminating human code writing and review. The core benefit is enabling custom, automated development pipelines with a strong foundation.

How It Works

Attractor is built upon NLSpecs, human-readable directives for coding agents. The repository outlines specifications for the Attractor agent, its agentic loop, and a unified LLM client. Users prompt compatible coding agents (e.g., Claude Code, Codex) with these NLSpecs, directing them to build the system as described at https://factory.strongdm.ai/. This methodology emphasizes "grown" software, where code is generated and validated through scenarios and a "Digital Twin Universe" (behavioral clones of third-party services) rather than traditional tests, aiming for "compounding correctness."

Quick Start & Requirements

Implementation requires an external, modern coding agent (e.g., Claude Code, Codex, OpenCode, Amp, Cursor). Users must provide the repository's NLSpecs as a prompt to this agent, instructing it to build Attractor based on specifications at https://factory.strongdm.ai/. Setup is specification-driven and relies on the user's chosen agent.

Highlighted Details

• Core principles: "Code must not be written by humans" and "Code must not be reviewed by humans."
• Utilizes "scenarios" (end-to-end user stories) and "satisfaction" (probabilistic validation) over traditional tests.
• Employs a "Digital Twin Universe" for validating agent behavior against behavioral clones of third-party services.
• Represents a shift towards "non-interactive development" and "grown software."

Maintenance & Community

No specific details regarding maintenance, contributors, community channels, or roadmap were present in the provided README snippet or the browsed URL.

Licensing & Compatibility

No licensing information or compatibility notes for commercial use were specified in the provided README snippet or the browsed URL.

Limitations & Caveats

This project is primarily a set of specifications and terminology (NLSpec) for building a system, not a deployable application. Successful implementation depends on the capabilities of the chosen external coding agent. Advanced concepts like the Digital Twin Universe are detailed on https://factory.strongdm.ai/. Economic feasibility is tied to significant token expenditure (e.g., "$1,000 on tokens today per human engineer").