Cover of Tweets From Thariq
tweets

Tweets From Thariq

@trq212 on Twitter

38 highlights
claude-agent-sdk agentic-tool-design agentic-design-patterns agentic-prompt-ui-design agentic-req-ux agentic-req-bee ai-use-case-scenario tool-idea-backlog

Highlights & Annotations

the typescript Claude Agent SDK v2 is honestly so good, and eventually we will likely migrate to it- if you’re using the typescript SDK I suggest you try it out asap! https://t.co/pVQQ7UnS1e (View Tweet)

Ref. AB37-A

here’s a hello world example if you want to check it out: https://t.co/nuFNHFeTUr (View Tweet)

Ref. 92FE-B

this is a great use of skills, creating a custom API that is LLM friendly

Ref. AA30-C

lots of alpha in making a plugin that teaches Claude Code how to be good at using your software https://t.co/drgJICe14z (View Tweet)

Ref. 051F-D

I’ve been helping my sister vibe code a WNBA stats site and have realized that 99% of what Claude Code says or displays isn’t useful to her because she’s not technical.

Ref. 798B-E

@AsphaltCowb0y made this into a plugin you can try: https://t.co/IQan4pZZjn https://t.co/41FydIGxmS

Ref. AD1F-F

for big features or new projects Claude might ask me 40+ questions and I end up with a much more detailed spec that I feel I had a lot of control over (View Tweet)

Ref. F247-G

the prompt I’ve been using is:

Ref. F85A-H

my favorite way to use Claude Code to build large features is spec based

Ref. B17C-I

@ImSh4yy Hey sorry turns out there was a bug where your displayed usage limits weren’t updating to show the 2x increase. Fix is rolling out now & you should see the correct percentage in the next hour or two (half of what it is right now)

Ref. 4865-J

one of the benefits of a detailed spec is that verification is also easier, after the work is done I like to add:

Ref. 16C0-K

Cowork is Claude Code packaged as a general agent for any task on your computer

Ref. 3B25-L

https://t.co/X2iu8WdIb8 (View Tweet)

Ref. AE6A-M

@tenobrus So skills, agents, etc. are a bit weird because they are not tools but instead parameters that the Skill and Agent tools can call with

Ref. 34A6-N

We’re rolling out Tool Search in Claude Code to reduce how much context MCP servers take up. https://t.co/KUU9YnCdeO (View Tweet)

Ref. 0C66-O

Our fix to remove flickering is rolled out again to everyone on Claude Code.

Ref. 861B-P

Slack in Cowork has saved me SO MUCH time

Ref. 030C-Q

Now in Claude Code: when you rewind a conversation using /rewind or hitting ESC twice, Claude can summarize the part of the conversation that was rewound.

Ref. 5299-R

@steveruizok Now add subway surfers (View Tweet)

Ref. 51F6-S

“I never see you at the club”

Ref. 25A1-T

agentic search and prompt caching rule everything around me (View Tweet)

Ref. 8D0A-U

Voice mode is rolling out now in Claude Code. It’s live for ~5% of users today, and will be ramping through the coming weeks.

Ref. 30ED-V

the team that made this are heroes, patriots https://t.co/kMgIhVvxnP (View Tweet)

Ref. CE48-W

in my culture we don’t say “I love you”. we say “I left a comment on your PR that’s a prompt you can just send straight to Claude Code” (View Tweet)

Ref. 1A7F-X

My favorite use case is to ask it to check error logs every few hours and create PRs for any actionable errors.

Ref. CEF2-Y

Today we’re launching local scheduled tasks in Claude Code desktop.

Ref. 5B42-Z

We just added /btw to Claude Code!

Ref. 908B-A

read more on setting up Code Review here:

Ref. E667-B

You can now configure Code Review to run manually on demand by typing @claude review in your Github PR or comments.

Ref. EC0D-C

A few end of week ships:

Ref. EE1D-D

I was just playing around to make something in @instance_11’s style btw, they’re the GOAT

Ref. 2136-E

the agent SDK is the easiest way to build agents

Ref. 063B-F

your agent should use a file system

Ref. 37E7-H

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

Ref. 3C7E-I

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

Ref. 4DB3-J

building an agent is more of an art than a science

Ref. BBD8-K

skills are the abstraction that all agents will build on

Ref. 35B6-L