{"slug": "i-built-an-open-protocol-to-make-ai-coding-agents-follow-senior-engineering", "title": "I built an open protocol to make AI coding agents follow senior-engineering workflows", "summary": "The article describes the creation of The Polyglot Protocol, an open-source framework designed to address common failure modes in AI coding agents, such as ignoring repository conventions, inventing APIs, and skipping validation. The protocol provides structured workflows including language-specific guidance, pre-codegen checklists, validation scripts, and post-codegen audit scoring to enforce senior-engineering discipline across polyglot codebases. It is model-agnostic, tested with various AI models, and released openly for community adaptation and feedback.", "body_md": "AI coding agents are getting better fast, but I kept running into the same failure modes:\n- they skip repository discovery\n- they invent APIs, flags, or config keys\n- they ignore existing project conventions\n- they add unnecessary infrastructure\n- they skip validation\n- they treat unsupported checks as silently done\n- they finish without a final codebase audit\nSo I built The Polyglot Protocol.\nIt is an open source senior-engineer workflow protocol for AI coding agents working across polyglot codebases.\nRepo:\nhttps://github.com/sabir-gbs/the-polyglot-protocol\nWhat it includes\n- guidance for 22 languages\n- language-selection rules\n- pre-codegen checklists\n- do-not-generate policies\n- validation scripts\n- adapters for Codex, Claude Code, and OpenCode\n- post-codegen audit scoring\nThe core idea\nBefore an agent changes code, it should:\n- inspect the repository\n- preserve existing conventions\n- choose the right language and tooling\n- avoid invented APIs or dependencies\n- validate the result\n- document unsupported checks as explicit\nN/A\n- finish with a final audit\nModel-agnostic by design\nI have tested the protocol with frontier models as well as models from Qwen, Grok, Kimi, MiniMax, and others.\nThe point is not to depend on one model. Strong models still benefit from stricter engineering discipline, and smaller/local models benefit from clearer constraints.\nWhy open source it?\nI want a practical, inspectable protocol that developers can adapt, critique, and improve.\nIf you use Codex, Claude Code, OpenCode, Cursor, Aider, or similar agents, I’d be interested in feedback:\n- Is the protocol too strict?\n- What guardrails are missing?\n- What should be optional vs required?\n- What would make this easier to adopt?", "url": "https://wpnews.pro/news/i-built-an-open-protocol-to-make-ai-coding-agents-follow-senior-engineering", "canonical_source": "https://dev.to/sabir_semerkant/i-built-an-open-protocol-to-make-ai-coding-agents-follow-senior-engineering-workflows-11ep", "published_at": "2026-05-24 00:11:31+00:00", "updated_at": "2026-05-24 00:32:24.552343+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "open-source", "developer-tools"], "entities": ["The Polyglot Protocol", "Codex", "Claude Code", "OpenCode", "Qwen", "Grok", "Kimi", "MiniMax"], "alternates": {"html": "https://wpnews.pro/news/i-built-an-open-protocol-to-make-ai-coding-agents-follow-senior-engineering", "markdown": "https://wpnews.pro/news/i-built-an-open-protocol-to-make-ai-coding-agents-follow-senior-engineering.md", "text": "https://wpnews.pro/news/i-built-an-open-protocol-to-make-ai-coding-agents-follow-senior-engineering.txt", "jsonld": "https://wpnews.pro/news/i-built-an-open-protocol-to-make-ai-coding-agents-follow-senior-engineering.jsonld"}}