I gave a general AI agent 27 minutes to be my candidate sourcing agent. What I really measured wasn’t time—it was attention.

inherent complexity of your task being higher than the agent's capabilities rather than some simple mistake that can be corrected.

A common mistake for people who are new to using agents is that they commit to making an interaction successful, even when an agent's work is veering off track…you should be ok discontinuing that conversation or manually taking over. Sending more messages is more likely a sign.

If using AI to aid in debugging, we recommend asking for a list of probable root causes rather than trying to debug and fix everything itself. Then, a human can decide based on their own experience which one is the real root cause.

Adding a new tool to render your Mermaid diagram. Everything is client side rendering, no data sent to any server. Your data is safe ;)

I think CodeX or any such "async" coding agent has a huge potential. I am optimistic to see the improvement and i am not holding back continuing trying these things out.

Then the whole process becomes counter productive. The goal i want is to materialize an sparking idea quickly without switching my main context to it. But if i have to keep telling it multiple times to do the job right. Then why am i still using it?.

I end up with multiple turns. And now i feel like i am spending more time on it then if i just quickly switch the context and use vs-code to do it with Copilot.

For example, i am asking it to create this page on my website to do Mermaid diagram rendering like this. I need to prompt it couple of times to get the resize right-resize the left hand side text editor and expect to see right hand panel resize accordingly

. to see the result can meet the requirements. Yes, I have already tuned on some agent prompts, and yes I have opened certain domain for Internet access. But even so, I could say I don't like the "taste" in code it produces.

The quality of the code it produces is definitely subpar to other coding agents i have been using. Due to its core interaction nature, every time i submit an idea i am not waiting and see it cooks but rather i check back and see its result. So the expectation is much higher. .

After several days usage #openaicodex my feeling is mixed. What I really enjoyed is its capability to work on multiple features async AND i can quickly turn my ideas into code even via ChatGPT mobile app. The UX is better than #CopilotCodingAgent. But. .

What a milestone, i finally got the Supabase integration done for my new website.

Just discovered that #Vercel allows you add a comment on a pre-production environment (link out from Github PR). I just wow! This is so much more useful than what I have to deal with day to day at my company.

I spend hours, even with help of ChatGPT, trying to figure out how to get NextJS hybrid app deployed on Azure Static Web App. While waiting for my next round of "deep research" result, i give it a try on #vercel, oh man it was smooth as a butter.

All this is one my persona toy project a developer tool box site for day to day use. Such as time convert, various encode/decode, Json manipulation, Emoji picker etc.

Huge 👏 to the GitHub & Microsoft teams—this is a power move for developer productivity. #AIEngineering #DevTools #GitHubCopilot.

🔒 Security kudos:.• Repo-level opt-in (can’t blanket-enable across org).• PR approver ≠ Issue creator → enforces peer review.• Workflows gated by manual approval.Microsoft clearly wore the “security hat” while shipping this.

📄 When the code compiles, Copilot updates the PR description—swapping the original plan for a detailed summary that’s ready for review. I can run staging deploys & smoke-test before merge, exactly like my normal flow. ✅.