The software development discipline where AI agents are orchestrated to autonomously plan, write, test, and evolve software based on human-defined goals and constraints.
- Team Adoption
- Code-Focused Agent Tools
- Talks & Conferences
- Case Studies and Practical Advice
- Standards
- Spec-Driven Development
- Multi-Agent Frameworks
- Browser, Web & Computer Use Agents
- Key Papers to Understand the Practice
Practical guidance for engineering teams.
- Claude Code Documentation - Anthropic's CLI agent for working on local codebases.
- OpenAI Codex Quickstart - OpenAI's coding agent: setup, API integration, and first code generation requests.
- Anthropic — Claude Code Best Practices - Prompting, tooling, and team-level workflows for Claude Code.
- OpenAI — A Practical Guide to Building Agents (PDF) - When to build an agent, design foundations, guardrails, and rollout.
- 12-Factor Agents - Principles for building reliable, production-grade LLM agents.
- Aider - AI pair programming in your terminal; supports complex multi-file refactors via agent-style planning.
- Cline - Autonomous coding agent for VS Code that can create/edit files, run commands, and use the browser with human-in-the-loop approval at each step.
- Continue - Open-source IDE extension (VS Code, JetBrains) for building, sharing, and running custom AI code assistants with any model.
- Goose - Block's open-source, on-machine AI agent that automates engineering tasks end-to-end with MCP-based tool extensions.
- OpenCode - CLI coding agent supporting 75+ LLM providers including local models, with MCP integration.
- OpenHands - Open-source Devin alternative; full software engineering agent with browser, terminal, and editor access.
- Plandex - Terminal-based AI coding engine designed for large, multi-file tasks.
- Roo Code - Autonomous AI coding agent for your editor; supports custom modes, multi-model providers, and orchestration of subtasks.
- Gemini CLI - Google's terminal agent; optimized for Gemini models with tool use and MCP support.
- Cursor - IDE built for agentic coding workflows; inline agent with codebase-wide context and multi-model support.
- Claude Code - Anthropic's agentic CLI that operates directly on your local codebase; optimized for Claude models.
- Codex - OpenAI's lightweight coding agent that runs in your terminal; optimized for OpenAI models.
Video content worth watching with your team.
- The Future of Coding is Agents — Andrej Karpathy (YC) - The trajectory from coding assistants to autonomous agents.
- Agentic Coding: The Future of Software Development with Agents — Armin Ronacher - Adopting Claude Code and agentic workflows in practice, by the creator of Flask and Sentry.
Real teams, real outcomes; learn from those ahead of you.
- My LLM Coding Workflow Going into 2026 — Addy Osmani - Specs-first planning, small iterations, and strong context and model selection in AI-assisted engineering.
- How Claude Code is Built - Claude Code's origins, architecture, and signals for AI-assisted software development.
- Building an AI-Native Engineering Team - Structuring and scaling engineering teams around AI-first workflows with Codex.
- everything-claude-code - Resources, tips, prompts, and examples for Claude Code.
- Anthropic — Building Effective Agents - Workflow and agent patterns for production agentic systems.
- agents.md - An open standard for guiding AI coding agents, giving them project-specific build, test, and code style instructions.
- MCP — Model Context Protocol - An open standard for connecting AI models to external tools, data sources, and services.
- MCP Reference Servers - Reference implementations of MCP servers.
- Awesome MCP Servers - Catalog of MCP server implementations for databases, APIs, browsers, and more.
Frameworks for turning specs and requirements into working software via agents.
- Spec Kit - GitHub's toolkit for spec-driven development, helping teams define what to build before letting agents implement it.
- BMAD-METHOD - Universal AI agent framework for agile, agentic-driven planning and execution across software, creative, and business domains.
Building blocks for orchestrating multiple agents with roles, memory, and tool access.
- LangChain - The most widely used framework for building LLM-powered applications, including multi-step agent workflows.
- LangGraph - Graph-based orchestration framework for building stateful, controllable multi-agent systems with explicit checkpoints and recovery.
- Microsoft AutoGen - Programming framework for building multi-agent AI systems with conversable agents that collaborate to solve tasks.
- CrewAI - Lean Python framework for orchestrating role-playing, autonomous AI agents that work together as a crew.
- Dify - Open-source platform for building LLM apps with agentic workflows, RAG, and observability; useful as the orchestration layer next to coding agents.
Agents that drive browsers, desktops, or the live web; useful for E2E testing, scraping, researching and verifying agent-built software.
- Browser Use - Make any LLM control a real browser; widely used for agentic E2E testing and automating web flows.
- Stagehand - Production-ready AI browser automation framework with natural-language actions on top of Playwright.
- Playwright - Microsoft's cross-browser automation library; the de-facto substrate for agent-driven browser testing and MCP browser servers.
- Open Interpreter - Lets LLMs run code locally with full system access; a foundational pattern for desktop-level coding agents.
- Firecrawl - Open-source web context API for agents; searches, scrapes, crawls, and converts live web content into clean Markdown or structured data, with MCP support for agent workflows.
- GLM-5: from Vibe Coding to Agentic Engineering - Technical report for GLM-5, a coding-focused model bridging casual vibe coding and structured agentic software engineering through advanced reasoning and tool use.
Contributions welcome! This list prioritizes practical, team-applicable resources over theoretical ones. When suggesting additions, please include a one-line description of why it is useful specifically to a team adopting agentic engineering.