Stop Closing the Door. Fix the House.

2026-03-17 · Source: AI & ML – Radar · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Intermediate, short

Summary

Open-source maintainers are increasingly frustrated by AI-generated pull requests (PRs) that often ignore conventions and contain "AI slop," leading some to stop accepting external contributions; however, the author advocates for adapting projects to responsibly integrate AI coding assistants instead of closing the door on contributors. Drawing from experience with the "goose" project, which has over 300 external contributors, the author outlines five key strategies: providing clear "HOWTOAI.md" and "AGENTS.md" guides for human and AI contributors, respectively, to set expectations and conventions. Further strategies include leveraging AI code reviewers with custom instructions to provide early feedback, maintaining a robust test suite as a critical safety net against bad AI-generated code, and automating quality checks via CI pipelines for linting, formatting, and project-specific issues like prompt injections. These measures aim to raise the quality bar and empower contributors to use AI tools responsibly, ensuring project quality without sacrificing the collaborative spirit of open source. The core message is to "fix the house" by preparing repositories for AI coding assistants rather than rejecting contributions.

Key takeaway

To manage the influx of AI-generated pull requests, open-source maintainers should proactively prepare their repositories by guiding both human contributors and AI agents. This involves establishing `HOWTOAI.md` for responsible human AI use, `AGENTS.md` to instruct AI agents on project conventions, and deploying AI code reviewers with custom instructions to pre-filter PRs for security and architectural correctness. These measures, combined with robust test suites and automated CI, reduce maintainer burden and enable high-quality contributions from a wider, AI-assisted community.

Topics

• Open-Source Maintenance
• AI Coding Assistants
• Automated Code Review
• Continuous Integration
• Software Testing

Code references

• block/goose

Best for: Software Engineer, MLOps Engineer, AI Engineer

Related on AIssential

• Code Is Not Cheap: How to Multiply Your AI’s Output With Software Fundamentals — AI-assisted coding, while prevalent, demands human engineering expertise to mitigate quality issues from "jagged" model outputs.
• AI agents in research: when productivity comes at the cost of apprenticeship — AI assistants like Claude Code and OpenClaw significantly enhance research efficiency across multiple tasks.
• Why Agents Still Need Humans — AI agents increase demand for human expertise by commoditizing routine tasks and creating an infinite backlog of new work.
• The Pulse: Industry leaders return to coding with AI — AI agents are re-engaging technical founders in hands-on coding, while AI-generated code copyright and competitive tactics face scrutiny.
• Stop Babysitting Agents, Start Authoring Outcomes — OpenProse transforms ephemeral AI agent chat sessions into reusable, reviewable, and versioned programs using logical English contracts.
• Dario Amodei, hype, AI safety, and the explosion of vibe-coded AI disasters — AI coding tools, while transformative, pose significant risks without expert human oversight and robust safety mechanisms.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI & ML – Radar.