Agentic software engineering

Agent skills and workflows

• Blueprint - source set of focused agent skills for designing, implementing, testing, reviewing, and shipping software changes.

Coding agents

• Codex CLI - source terminal agent with sandbox and approval controls.
• Gemini CLI - source terminal agent built around Gemini and extensible tools.
• Cursor CLI
• Aider - source pair programmer with Git integration and broad model support.
• OpenCode - source, provider-independent terminal agent with a client-server architecture.
• Cline - source coding agent available as an editor extension, CLI, and SDK.
• Continue - source coding agents for IDE and CI workflows with source-controlled configuration.
• GitHub Copilot coding agent
• Neo - source, workflow-first terminal coding agent with subagents, skills, sandboxed tools, and multiple model providers.
• Build Your Own Coding Agent Like Pi

Deployment and observability

• SWE-bench - Bench](https://www.tbench.ai/leaderboard/terminal-bench/2.1).

Software factories and agent orchestration

• Harness Engineering
• Codex Orchestration with Symphony - agent runs.
• How We Built Our Multi-Agent Research System - worker agents, parallel search, evaluation, and operational reliability.
• Factory - preview control plane for scheduling and coordinating Pi, Codex, and Claude Code workers across Git repositories.

Build AI systems

Agent frameworks

• Google Agent Development Kit
• Pydantic AI
• LangGraph - level orchestration for long-running, stateful agents.
• OpenAI Agents SDK
• Microsoft Agent Framework - based workflows.

Deployment and observability

• Langfuse
• LiteLLM
• vLLM

Durable and asynchronous agents

• Effective Harnesses for Long-Running Agents
• Running Agents - execution patterns in the OpenAI Agents SDK.
• Human-in-the-Loop
• Gemini and Temporal Durable Agent

Evals and reliability

• OpenAI Evals - source framework and registry for evaluating language models and systems.
• Demystifying Evals for AI Agents
• Promptfoo - team checks for LLM applications.

Guides and playbooks

• Building Effective Agents
• A Practical Guide to Building Agents
• Awesome DeepSeek Agent

LLM application engineering

• OpenAI Cookbook
• Anthropic Prompt Engineering
• Effective Context Engineering for AI Agents - running agents.
• 12-Factor Agents
• OWASP Top 10 for LLM Applications

Protocols and interoperability

• Model Context Protocol
• Agent2Agent Protocol - neutral specification for agent discovery, task delegation, streaming, asynchronous updates, and cross-platform communication.

Retrieval and data

• Docling
• Haystack

Contributing

Software factories and agent orchestration

• AI Engineer YouTube channel

Learn

Books

• Designing Machine Learning Systems
• AI Engineering
• The 100-Page Language Models Book
• Artificial Intelligence: A Modern Approach
• Deep Learning
• Reinforcement Learning: An Introduction
• Build a Large Language Model from Scratch
• Hands-On Large Language Models
• LLM Engineer's Handbook - tuning, serving, and production workflows.
• Deep Learning: Foundations and Concepts - grounded treatment by Christopher and Hugh Bishop.
• Understanding Deep Learning
• Speech and Language Processing
• Machine Learning Bookcamp - based introduction to building and deploying machine learning systems by Alexey Grigorev.

Courses

• Google Generative AI Learning Path
• Hugging Face LLM Course - tuning, datasets, and modern NLP tooling.
• Fast.ai Practical Deep Learning - first introduction to deep learning.
• Full Stack Deep Learning
• MIT 6.S191: Introduction to Deep Learning
• DeepLearning.AI Short Courses
• Stanford CS336: Language Modeling from Scratch
• Made With ML
• AI Engineer

Foundational papers

• Attention Is All You Need
• Scaling Laws for Neural Language Models
• Language Models are Few-Shot Learners - context learning at scale.
• Constitutional AI
• Training Compute-Optimal Large Language Models
• Retrieval-Augmented Generation - intensive tasks.
• LoRA - rank adaptation for parameter-efficient model fine-tuning.
• Training Language Models to Follow Instructions with Human Feedback
• ReAct - using language-model agents.
• Direct Preference Optimization