AGENTS.md for OpenAI Codex, Cursor, Google Jules

Today’s top AI Highlights:

1. OpenAI’s standardized README for AI coding agents

2. Full-stack implementation code from a research paper in one click

3. Vibe build voice AI agents with RAG and background reasoning

4. Microsoft’s free interactive Python tutorial on MCP servers

5. Project management system for Claude Code

& so much more!

Read time: 3 mins

AI Tutorial

Building targeted B2B outreach campaigns is one of the most time-consuming aspects of sales and marketing. The challenge isn't just finding companies; it's discovering the right decision-makers, researching genuine insights, and crafting personalized messages that actually get responses.

In this tutorial, we'll build a multi-agent AI email outreach system using OpenAI GPT-5, Agno for orchestrating agents, and Exa AI for intelligent web search. This system automates the entire outreach pipeline - from company discovery to personalized email generation - delivering professional, research-backed outreach emails in minutes instead of hours.

Our multi-agent system conducts real research on each company using website content and Reddit discussions and ensures every email feels genuinely personalized.

We share hands-on tutorials like this every week, designed to help you stay ahead in the world of AI. If you're serious about leveling up your AI skills and staying ahead of the curve, subscribe now and be the first to access our latest tutorials.

Don’t forget to share this newsletter on your social channels and tag Unwind AI (X, LinkedIn, Threads) to support us!

Latest Developments

Research Paper → Full-Stack Code in One Click 🔬📃 🖥️

This team of AI agents can not only read research papers but also turn them into production-ready code in minutes.

DeepCode is an opensource multi-agent system that automates the entire journey from research papers to production-ready implementations. You just need to give it the research paper, that’s it!

The system handles everything from extracting algorithmic logic from academic documents to generating complete codebases with tests and documentation. DeepCode specializes in translating complex mathematical models and research methodologies into functional software. It coordinates multiple specialized agents that handle document parsing, code planning, reference mining, and implementation generation using MCP tools.

Key Highlights:

1. Research-to-Production - Automatically extracts algorithmic logic and mathematical models from academic papers, then generates optimized implementations while preserving computational complexities.

2. Multi-Agent Team - Eight specialized agents work together through intelligent orchestration, handling everything from document parsing to code generation and quality assurance.

3. CodeRAG System - Discovers relevant repositories and frameworks through search algorithms, building comprehensive knowledge graphs of codebases for globally-aware code recommendations.

4. Complete Automation - Generates not just code but entire project structures including database schemas, API endpoints, frontend components, test suites, and comprehensive documentation.

One Simple File Format for README for Coding Agents 🦮🤖

Every AI coding agent speaks a different language when it comes to project instructions.

Claude Code expects a CLAUDE.md file, Cursor looks for .cursorrules, Google Jules reads AGENTS.md, and the list keeps growing as new agents launch. You either stick to one tool, or have to maintain multiple instruction files that essentially contain the same information about your codebase.

Now, a collaborative initiative between OpenAI, Google, Cursor, AmpCode, FactoryAI, and Roo Code aims to fix this mess with AGENTS.md - a single, standardized format that works across every major coding agent. This simple format lets you write your project context, build commands, and coding standards once, then use them with any agent you choose. The goal is interoperability: one file that OpenAI Codex, Google Jules, Cursor, and future agents can all understand natively.

Key Highlights:

1. Agent-agnostic design - Replaces tool-specific files with a single format that works across OpenAI Codex, Google Jules, Cursor, and other participating agents (hoping Anthropic joins the list too for Claude Code).

2. Project context - Contains build commands, testing instructions, code style guidelines, security considerations, deployment steps, and project-specific conventions that agents need to work effectively.

3. Nested configuration - Handles complex monorepos through hierarchical AGENTS.md files where subproject-specific instructions take precedence over general project rules.

4. Adoption - Backed by major companies including OpenAI, Google, Cursor, AmpCode, FactoryAI, and Roo Code, with over 20,000 open-source projects already implementing the standard.

You Don’t Need to Be Technical. Just Informed

AI isn’t optional anymore—but coding isn’t required.

The AI Report gives business leaders the edge with daily insights, use cases, and implementation guides across ops, sales, and strategy.

Trusted by professionals at Google, OpenAI, and Microsoft.

👉 Get the newsletter and make smarter AI decisions.

Lead with AI—no coding needed

Quick Bites

Google just launched URL context in the Gemini API. This feature lets the model directly access webpages, PDFs, and images by simply including URLs in requests. When you provide a URL, the tool first attempts to fetch the content from an internal index cache. If a URL is not available in the index (for example, if it's a very new page), the tool automatically falls back to do a live fetch. This helps balance speed, cost, and access to fresh data.

Microsoft has released a free and interactive Python tutorial for MCP servers, to create AI agent applications that can access external tools and data sources. The hands-on guide walks through building two projects: a Python learning companion that generates personalized coding challenges, and an AI research discovery service that finds trending papers and creates study plans. The tutorial takes 15-75 minutes total and includes everything from basic MCP concepts to advanced implementations.

Voice AI platform Cartesia launched Line, a platform to build, test, and deploy voice AI agents with features like background reasoning, RAG, and ultra-low latency STT and TTS models. It combines the best of both vibe coding and code-first worlds - you can start with a simple prompt to generate the code for your agent, export to GitHub or CLI, and then work further on it to enhance the agent’s capabilities. The platform also lets you test and deploy the agents in just a few clicks. Live now with free credits for early users.

Tools of the Trade

1. CCPM - A project management system that turns PRDs into GitHub issues and lets multiple Claude Code agents work on different parts of a project simultaneously. It keeps track of what each agent is doing so you don't lose context when running parallel development tasks.

2. Cipher - Opensource self-improving AI memory layer that auto-creates and retrieves coding memories that scale with your codebase. Works across Cursor, Windsurf, Claude Code, Gemini CLI, and other agents via MCP. Has a dual memory layer where System 1 captures programming concepts and business logic, and System 2 captures reasoning steps of the model when generating code.

3. Parse - AI web scraping platform that automatically extracts data from websites in seconds. It handles dynamic JavaScript content and exports data in multiple formats, including JSON and CSV.

4. Awesome LLM Apps: A curated collection of LLM apps with RAG, AI Agents, multi-agent teams, MCP, voice agents, and more. The apps use models from OpenAI, Anthropic, Google, and open-source models like DeepSeek, Qwen, and Llama that you can run locally on your computer.
   (Accepting GitHub sponsorships now)

Hot Takes

1. Hot take: AI doesn’t make you dumb, if you think it does, that says more about you than AI. ~
   Ashutosh Shrivastava
   
   

2. We’ll know AGI is postponed once the labs start building features to choose our own background colors and fonts. ~
   Harj Taggar

That’s all for today! See you tomorrow with more such AI-filled content.

Don’t forget to share this newsletter on your social channels and tag Unwind AI to support us!

Unwind AI - X | LinkedIn | Threads

Awesome LLM Apps | Sponsor Us

PS: We curate this AI newsletter every day for FREE, your support is what keeps us going. If you find value in what you read, share it with at least one, two (or 20) of your friends 😉