I've been seeing a shift in how agents use tools.
CLIs + Skills seem like the new pattern for giving agents access to do real work.
A lot of people are moving away from MCPs to this.
Made another Sale for Formaly.io!
Every Spike in this graph motivates me even more!
Shipping some cool features there!
Large MoE models break latency budgets in production 👀
Not because of hardware. Because teams optimize for throughput instead of tail latency.
P90/P99 is what kills products.
Speculative decoding isn't an optimization. It's a prerequisite for real SLAs.
Full breakdown by @demian_ai 👇
You wouldn't run a backend service without monitoring.
Why treat AI inference differently?
@nebiustf gives you:
• Traffic volume + throughput
• TTFT tracking
• Error breakdowns
• Prompt size distributions
All in near-real time. MLOps hygiene done right.
You wouldn't run a backend service without monitoring.
Why treat AI inference differently?
@nebiustf gives you:
• Traffic volume + throughput
• TTFT tracking
• Error breakdowns
• Prompt size distributions
All in near-real-time. MLOps hygiene done right.
More tools need to expose MCP Logs like this 👀
As more agents start using MCP servers, visibility into what's happening will be very important
This should be the new standard.
Found on @InsForge_dev dashboard.
Deployed my First Site on @appwrite Sites!
Dashboard interface displaying authentication options, data for contacts and interactions, with enabled and disabled OAuth providers.
Currently Playing with @InsForge_dev
and really liked their Visualizer. (Apart from the fact that, I can manage auth, Db, models, storage, and Deployment all from CC)
Will make a detailed video on it soon
Found an infographic for the Claude Code Agent Skills section 👀
Started using it today, and it's pretty solid for task consistency.
If you're using Skills differently or have tips, drop them below!
Would love to learn how others are using this.
These emails from @dodopayments hit different 👀
Pure dopamine every time a sale notification drops.
First @ProductHunt launch just wrapped up!
Formaly hit #3 on the leaderboard 👀
40+ comments and feedback, tons of support, and got my first paying customer!
Thanks everyone who showed up and voted!
A Claude Code optimization culture is forming 👀
• Break work into tasks, not mega-prompts
• Run parallel agents with clear roles
• Let Claude generate, use GPT-class models to review
• Treat agents like processes, not chat windows
The real skill now isn’t prompting. It’s agent orchestration.
Made my first SaaS income this week 👀
$26 so far.
Feels good. Building more.
Page visits on Formaly went from 0 to 100 after the @ProductHunt launch 👀
The PH effect is real.
A lot of people asked me the what, how, and why behind Formaly.
So I wrote a short blog breaking it all down, the idea, the decisions, and where it’s headed.
👉 formaly.io/blog/introd...
Hope this clears things up and gives some context.
Claude Code /insight Feature is super helpful!
It went through all my previous interactions and gave a detailed report:
- Tools used the most
- Session Types
- My Claude usage style
- What I do right and wrong
- Which features I should use
Most importantly how can I improve my workflow
We're at #3 on @ProductHunt right now 👀
Can we hit #1?
Would mean a lot if you could support us
90% off all Formaly tiers for the ProductHunt launch 👀
The first tier is under $1.
Grab it while it lasts!
x.com/Arindam_172...
I'll be attending AI Impact Summit next week 👀
Excited to meet everyone building cool stuff in the space.
If you're around, let's meet!
All Set for My First @ProductHunt launch!😼
Wish me Luck!
Got rate-limited on @cursor_ai while using Opus 4.6!
Was working on my project and hit rate limits on Cursor (pro plan). Auto models aren't my thing.
My fix: plugged in @Kimi_Moonshot K2.5 via @nebiustf
Works surprisingly well. Drop-in replacement for Opus 4.5
x.com/nebiustf/st...
This is how I run an SEO audit with Claude these days 👀
• Plug in the SEO-audit skill from skills.sh
• Ask Claude for a detailed audit report
• Sit back and let it do its thing
Way faster than manual checklists.
Here’s how I did it 👇
Recently found the effort level setting in Claude Code 👀
And I’ve been using it a lot lately.
For lighter tasks, you can use low effort and save tokens.
Clean way to manage usage efficiently
x.com/bcherny/sta...
