CloudAI-X
@cloudxdev
Followers
3K
Following
9K
Media
203
Statuses
2K
.jsx .js .tsx .ts [email protected]
Joined November 2020
I have published the whole agents, skills, commands, hooks files that I use as a "project-starter". Check it out https://t.co/1fOMe2jOeL
github.com
Universal Claude Code workflow plugin with agents, skills, hooks, and commands - CloudAI-X/claude-workflow-v2
17
70
767
Man I have soooo many articles to read and bookmarks to visit I don't know how to overcome it lol And also I am working/reading/learning nonstop all day, there is just so many new things
0
0
4
Just tried @vercel_dev new bash-tool with @aisdk v6. 1- Create a virtual filesystem 2- Give model(Haiku in this case) a bash tool, and it runs find/cat/grep to explore your code. Zero token bloat. The agent pulls only what it needs. Output: export const main = () =>
0
0
2
"Bash is all you Need" "Bash is what makes Claude Code so good." Thats what @trq212 said on his Claude Agent SDK Workshop. So I tend to listen to him, will dipe into bash tools more deeply, if you are interested here is a free resource: https://t.co/lquVdn513E
2
0
8
Peak performance Thanks to @dejavucoder for sharing a great blog post Must read: https://t.co/8HeWUeOYaa
2
3
52
GM There is something clearly different with Claude Code. Much faster, direct and brief. What secret sauce they added again?
0
0
1
Don't you dare shout at me claude code I have been baby sitting you ALL DAY
1
0
2
Published an @xai cookbook. Main docs site was lack of it so made my own, will add more comprehensive examples/tools @grok 4.1-fast is pretty efficient btw Play around and share your thoughts https://t.co/yPiiAv9iwu
github.com
Contribute to CloudAI-X/xai-cookbook development by creating an account on GitHub.
1
1
14
Huge Optimization From @warpdotdev They've shipped an optimized MCP Search Subagent that intelligently finds/loads only the relevant tools from MCP servers 1- Up to 26% token savings on MCP-heavy tasks, 10% overall 2- Subagent does natural-language search in an isolated context
MCP servers can bloat your context window, so we built a solution: the MCP search subagent. This tool reduces context usage by up to 26% in our testing, and works with *any* AI model. And we had had product engineer Matthew Albright explain how we built it (in thread 🧵)
0
0
1
GM @SuperGrok Heavy gives you 50$ worth credit for API every month I also find @grok 4.1-thinking evolved a lot Good job @xai
3
0
3