superthread of every resource I have published about advanced claude code usage, research-plan-implement, and advanced context engineering for coding agents 👇 (mostly chronological)
august 20th - 14 minute talk @ YC https://www.youtube.com/watch?...
Aug 5th - 75min livestream - one-shotting fixes to a 300k LoC rust codebase https://www.youtube.com/watch?...
Sept 21 - getting AI to work in complex codebases - blog post https://github.com/humanlayer/...
Sept 23 - HN thread on the sept 21 post https://news.ycombinator.com/i...
Oct 14th - 3 hour hands-on shipping OSS PRs with research-plan-implement https://x.com/dexhorthy/status...
Aug 25th - we put a coding agent in a while loop and it shipped 6 repos overnight https://github.com/repomirrorh...
Oct 28th - Running Agents in an Infinite loop w/ @GeoffreyHuntley 's "ralph wiggum" technique https://x.com/dexhorthy/status...
@dexhorthy Thanks a ton man
@dexhorthy @threadreaderapp unroll
@dexhorthy Excellent resource compilation on advanced context engineering! The research-plan-implement framework is crucial for agent reliability. Your layered approach helps developers structure complex coding tasks effectively. Keep sharing!
This is a fascinating way of working with multiple coding agents at once: "Before doing anything else, read ALL of AGENTS dot md and register with agent mail and introduce yourself to the other agents. Then coordinate on the remaining tasks"
claude skills have completely replaced dotfiles for me. getting to focus on actual coding instead of endless ricing and config tinkering makes the existential dread of daily life feel just a bit lighter
If you're setting up your Coding CLI, here are some tools worth adding. I've found fd, ripgrep, and ast-grep especially useful when working with codex.
nano-vLLM is pretty cool lightweight vLLM implementation (it’s actually a side project by a deepseek engineer!)
