@cojobrien Take a picture of that historic building. Build a model of it. Put it in a book and museum. Then tear it down and build something better. *yes better being the key word
The other legitimately useful tool here is for my newsletter blurb I do each Friday. The system knows what articles were posted this week, what the contents were and the authors. Then I add my own notes and flare and Eli and @grok give me 3 options to choose from. (or I keep editing)
Just loving it all.
I know I already showed this off but man this is one of the better systems for myself I've built in a while.
Looking at a @txGarage submission from author Harold Allen. I just add the submission and my Eli agent-harness knows who Harold is, knows his voice, knows the people from @Ford and other places worth reaching out to for sharing, and knows where I plan on sharing and promoting the article. Then sends all that to @Grok who builds out all this great content for me in one go!
Just a great workflow and it's only one small part of the full system.
Open-sourced Harness 101 β a clean, minimal TypeScript agent harness with a full step-by-step tutorial.
Includes planner/implementer/reviewer profiles, memory compaction, sub-agents, safe tools, and more.
If you're building agents and want to understand the wiring, check it out:
https://t.co/i89PKwYy4f
Would really appreciate a β if you find it useful!
Just published a new field note: Harness 101 β a clean TypeScript agent harness with a detailed step-by-step breakdown.
If you want to actually understand how agentic systems work (profiles, memory compaction, sub-agents, etc.), this oneβs for you.
https://t.co/PIyLy2lNNt