In Cline, I can...

2. /newtask to continue the current to a new task
docs.cline.bot/features/sla...

3. When Cline edits a file. and before I approve, Cline stops am waits..I can edit the file and then Approve the entire change. Cline sees my my User Edits and only then continues. ❤️

Thanks all🙏

After trying Codex and others....Codex was hands down the winner.

The project: To port this site written in Svelte to my Flowbite Blazor library
Svelte Admin Panel: flowbite-svelte.com/admin-dashbo...
Blazor Admin Panel: flowbite-blazor-admin-dashboard.pages.dev

Codex was about 75% of this

I also recommend using Codex Vs Code Extension + Microsoft Playwright MCP.

It can then run and interrogate the running code as well as take screenshots for verification.

But I like the UX better in the IDE as it's easier to see what is going on as well as provide your own screenshots

If this is what it takes to actually get good music to be played more, I am all in. From a musical standpoint, this has a lot going on as well as sensible lyrics. Noice.

For example github.com/cline/cline/...

This is essentially now supported by Copilot, Claude Code, and Gemini CLI

I would still 100% put an MCP server in front of my enterprise service that supports Auth🔐

With orgs finally undstanding the power of custom commands (i.e. workflows) ...and now wanting centralized management..

Remote MCP prompts solve this but tool makers are myopically focused on MCP tools 🙃

@simomo.bsky.social interesting read as always.

MCP 100% has its role

As quoted by Anthropic
"Claude.ai does not currently support centralized admin management or org-wide distribution of custom Skills."

MCP Prompts are, in fact, remote skills. If we just had MORE AI tools support them 😢

This is the way.

Unfortunately social media and unregulated media outlets have already poisoned the critical thinking skills and increased tribalism

Gemini has been AB testing Gemini 3 and open AI studio recently. Something like 1 in 50 requests will get you a response from the new model. YouTuber painstakingly put this through his benchmarks and the results were astounding. Especially coding

Next...create a custom command (i.e. workflow) that leverages sub-agents for parallel data gathering or parallel updating 🤯

Man, I am struggling with #copilot complexity.

I repeated the same test and it got first shot 👌

Looks very intriguing actually. To what extent will the application of the linear algebra requirement be practically applied during this program?

..however this --can-- be overcome. A human maintained spec and test plan, etc one has prayer...but only slightly