Reading List on Agentic Coding

This is live reading list LLMs, AI and agentic coding. It will be updated as I discover more material.

Entries without description / naked links are the ones I have not read in detail just yet.

Important Ideas

• Make agents with different models.
  Make agents with different LLMs: one agent writes code, another agent reviews and/or tests code. Otherwise LLMs tend to agree with themselves. Source
• Agents and models have different strenghts and weaknesses, lean into that.

General

• Levels of Agentic Engineering — Overview of levels of adoption of agentic engineering.
• https://nanonets.com/cookbooks/structured-llm-outputs/
• https://medium.com/@sampan090611/tensorflow-is-dead-pytorch-won-2d0bc6e9b1a4
• https://oxide-and-friends.transistor.fm/episodes/engineering-rigor-in-the-llm-age
• https://justsitandgrin.im/posts/coding-agents-use-cases/
• AI Adoption Journey
  • https://news.ycombinator.com/item?id=46903558
  • https://mitchellh.com/writing/my-ai-adoption-journey

Theory

• https://www.confident-ai.com/blog/llm-evaluation-metrics-everything-you-need-for-llm-evaluation
• Visualising Neural Networks
  • https://bbycroft.net/llm
  • https://news.ycombinator.com/item?id=46871654
  • https://visualrambling.space/neural-network/ Visualizing LLMs
  • https://r2d3.us/visual-intro-to-machine-learning-part-1/
  • https://sebastianraschka.com/llm-architecture-gallery/
• Understanding AI agents
  • https://news.ycombinator.com/item?id=46545620
  • https://www.mihaileric.com/The-Emperor-Has-No-Clothes/
  • https://fly.io/blog/everyone-write-an-agent/

Advanced / Teams of Agents

• Welcome to Gas Town
• Gas Town
• https://lobste.rs/s/txknsm/my_thoughts_on_gas_town_after_10_000_hours
• Paper - LLM-Based Multi-Agent Systems in Software Engineering
• Advanced Context Engineering for Coding Agents⭐

Cookbooks and Examples

• https://github.com/github/awesome-copilot
• https://docs.github.com/en/copilot/tutorials/copilot-chat-cookbook
• https://heidenstedt.org/posts/2026/how-to-effectively-write-quality-code-with-ai/
• https://github.com/wesammustafa/Claude-Code-Everything-You-Need-to-Know
• https://github.com/shanraisshan/claude-code-best-practice
• https://github.com/punkpeye/awesome-mcp-servers
• Agentic Patterns
  • https://github.com/nibzard/awesome-agentic-patterns
  • https://www.agentic-patterns.com/
• https://github.com/ChrisWiles/claude-code-showcase?tab=readme-ov-file
• Advanced - Desplega.ai
  • https://news.ycombinator.com/item?id=47165046
  • https://github.com/desplega-ai/advanced-context-engineering-for-coding-agents/blob/main/ace-fca.md
  • https://github.com/desplega-ai/agent-swarm
• https://simonwillison.net/guides/agentic-engineering-patterns/prompts/

Code Examples

• [Using Google Gemma multi-modal model for understanding speech and video locally] https://github.com/fikrikarim/parlor

Interesting Tools

• Superpowers⭐
  Agentic coding framework defining agents, skills and patterns for efficient multi-agent development.
• Get Shit Done⭐
  Opinionated light-weight agentic spec-driven meta-prompting coding system that poses itself as simpler alternative to SpecKit. Also see overview here. And see video here.
• Herdr⭐ Terminal multiplexer for coding agents
• OpenCode⭐
  FOSS harness for agentic CLI coding similar to Claude Code, Copilot CLI etc.
• Pi
  Minimalistic FOSS harness for agentic CLI coding.
• https://entire.io/vision
• https://developer.microsoft.com/blog/making-windows-terminal-awesome-with-github-copilot-cli
• https://openclaw.ai/

Management and Psychology

• How Is AI-assisted Coding Going for You?
  Mix of personal opinions ranging from happy to full on doom, very little to almost everything. It's pretty much "the future is already here, it's just unevenly distributed. One thing almost everyone agrees on - there is big shift underway.
• The machines are fine. I'm worried about us.⭐
  Very good article about using LLMs in academia and the learning gaps they create. Principled and nuanced article.
• Dark Flow
  The author draws parallels between using LLMs and gambling - both are feeding users losses disguised as partial success to create a psychological state of "dark flow". The conclusion is that LLMs may be making us less productive in the long term and causing skill atrophy. So creativity and thinking still matter.
• https://shumer.dev/something-big-is-happening
• The rise of industrial software
• Future of software development conference
  • https://martinfowler.com/fragments/2026-02-18.html
  • https://annievella.com/posts/finding-comfort-in-the-uncertainty/
  • https://www.thoughtworks.com/content/dam/thoughtworks/documents/report/tw_future%20_of_software_development_retreat_%20key_takeaways.pdf
• https://addyosmani.com/blog/next-two-years/

Futuristic Musings

• Singularity will occur on Tuesday
  • https://news.ycombinator.com/item?id=46962996
  • https://campedersen.com/singularity
• Your AI Super Doctor Will See You Now - On AI medicine of the future.
• https://dlants.me/agi-not-imminent.html
• https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/
• https://martinalderson.com/posts/wall-street-lost-285-billion-because-of-13-markdown-files/
• https://martinalderson.com/posts/ai-agents-are-starting-to-eat-saas/

Unsorted Queue

• https://neilkakkar.com/productive-with-claude-code.html

• https://www.claudecodecamp.com/p/i-m-building-agents-that-run-while-i-sleep

• https://www.scientificamerican.com/article/why-developers-using-ai-are-working-longer-hours/

• https://www.reddit.com/r/ExperiencedDevs/comments/1rnkv2t/the_ai_coding_productivity_data_is_in_and_its_not/

• https://andychen32.substack.com/p/the-enterprise-context-layer

• https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-deep-think/

• https://pub.spillwave.com/agentic-coding-gsd-vs-spec-kit-vs-openspec-vs-taskmaster-ai-where-sdd-tools-diverge-0414dcb97e46

• https://blog.information-superhighway.net/on-the-need-for-understanding

• https://www.apideck.com/blog/mcp-server-eating-context-window-cli-alternative

• https://simonwillison.net/guides/agentic-engineering-patterns/prompts/

• https://arxiv.org/html/2404.04834v4

• https://news.ycombinator.com/item?id=47394022

• https://kanfa.macbudkowski.com/vibecoding-cryptosaurus

• https://chrlschn.dev/blog/2026/03/mcp-is-dead-long-live-mcp/#summary

• https://modelcontextprotocol.io/docs/getting-started/intro

• https://learningloom.substack.com/p/the-future-belongs-to-those-who-can

• https://www.bhusalmanish.com.np/blog/posts/why-claude-wins-coding.html

• https://www.thoughtworks.com/insights/blog/agile-engineering-practices/where-does-the-rigor-go

• https://stevekrouse.com/precision

• https://www.thoughtworks.com/about-us/events/the-future-of-software-development

• https://composio.dev/content/openclaw-security-and-vulnerabilities

• https://contextbento.substack.com/p/openclaw-high-level-architecture

• https://blog.dailydoseofds.com/p/anatomy-of-the-claude-folder

• https://github.com/facebookresearch/hyperagents

• https://github.com/axeldelafosse/loop

• https://axeldelafosse.com/blog/agent-to-agent-pair-programming

• https://lab.puga.com.br/cog/#/

• https://github.com/itigges22/ATLAS

• https://github.com/garrytan/gstack

• https://neilkakkar.com/productive-with-claude-code.html

• https://developer.chrome.com/blog/chrome-devtools-mcp-debug-your-browser-session

• https://www.reddit.com/r/ClaudeCode/comments/1rw9vsi/pro_tip_just_ask_claude_to_enable_playwright/

• https://standupforme.app/blog/some-uncomfortable-truths-about-ai-coding-agents/

• https://mariozechner.at/posts/2026-03-25-thoughts-on-slowing-the-fuck-down/

Non-AI Temp

• https://medium.com/@rafael-22/the-clean-architecture-i-wish-someone-had-explained-to-me-dcc1572dbeac
• https://learn.microsoft.com/en-us/dotnet/architecture/microservices/