what i did

1. created a metacognition skill for claude. runs at the end of each session to:
   1. identify learnings/mistakes
   2. update .md files accordingly so it doesn't make the same mistakes in future sessions
2. asked claude for feedback on website SEO - executed on those changes.
3. added sandbox mode so claude doesn't rm-rf my home directory
4. worked with claude to create a development plan for a simple promptLibrary CRUD app i'm working on.
   1. for learning purposes i manually made changes on project to convert to typescript and switch from commonjs to esmodules
5. added relevant learnings to mochi spaced repetition system
6. read about harnesses in the monorepo

what i learned using claude:

1. claude can teach me skills i don't know.
   1. i.e. realized i had some gaps in typescript knowledge, asked claude questions.
2. i need to inhibit more of the traits i infuse in claude
   1. i caught myself asking claude to think in ways i rarely do.
   2. i need to:
      1. take a step back and ask myself - 'what makes a good website design?' before just immediately designing something.
      2. have a stronger ooda loop - always identify areas of improvement, how i can be better tomorrow.
3. asking claude to let me know where its making assumptions helps avoid pitfalls it would have made otherwise.
4. excited by the different ways we can work with these agents to make them even better!
   1. metacognition, etc.
5. need to get more comfortable multitasking. while an agent is running on a longer task - do something else.