Skip to content
Thariq@trq212 · Mar 21, 2026

I put a lot of heart into my technical writing, I hope it's useful to you all. 📌 Here's a pinned thread of everything…

13 tweets2 min read♥ 8.6Koriginal

Summary

A developer shares a curated collection of their technical writing on AI agents, covering key topics like skills, prompt caching, agent SDKs, bash scripting, and agentic coding practices. Most posts will be republished on the Claude blog.

Summarized by ThreadOut AI from the full thread. May miss nuance — read the thread below.

  1. #1

    I put a lot of heart into my technical writing, I hope it's useful to you all. 📌 Here's a pinned thread of everything I've written. (much of this will be posted on the Claude blog soon as well)

  2. #2

    skills are the abstraction that all agents will build on

  3. #3

    building an agent is more of an art than a science

  4. #4

    imo my highest alpha post is on prompt caching, but it's only really relevant if you're building agents from scratch

  5. #5

    playgrounds are one of the best ways to iterate on ideas visually

  6. #6

    your agent should use a file system

    Thariq@trq212 · Sep 22, 2025

    Your Agent should use a File System This is a hill I will die on. Every agent can use a file system. The file system is an elegant way of representing state that your agent could read into context & allowing it to verify its work. 🧵on why and examples

  7. #7

    bash is all you need

    Thariq@trq212 · Oct 27, 2025

    Why even non-coding agents need bash I've done dozens of calls with companies making general agents over the past few weeks and my advice generally boils down to: "use the bash tool more" Here's a concrete example from my email agent:

  8. #8

    the agent SDK is the easiest way to build agents claude.com/blog/building-…

  9. #9

    been meaning to write more longform on specs, but in the meantime this post is still good

    Thariq@trq212 · Dec 28, 2025

    my favorite way to use Claude Code to build large features is spec based start with a minimal spec or prompt and ask Claude to interview you using the AskUserQuestionTool then make a new session to execute the spec

  10. #10

    you have to think more carefully about session management with one million context, not less

  11. #11

    I've been using HTML for planning, speccing, exploration, code review, reports and a lot more.

  12. #12

    How dynamic workflows allow Claude Code to handle whole new types of tasks

  13. #13

    one of the core skills of agentic coding is reducing your unknowns