MadaShindeInai

@rauchg conakranie.vercel.app Opus4.6/Sonnet4.6

Show me the thing you’ve built with AI you’re most proud of. Reply with a working product URL and what model / agent you primarily used.

@dshukertjr @supabase Display resume button for stopped project in dashboard, when you open it from mobile view.

What would you add to Supabase if you could wave a magic wound?

Running Claude Code with --dangerously-skip-permissions on your host? You shouldn't. I containerized mine without losing DX. Docker Sandboxes (sbx) + a tiny shell function gets you most of the way there. Full writeup: @madashindeinai/sandboxing-claude-code-a-nearly-frictionless-setup-for-daily-use-a50f557ab33a" target="_blank" rel="nofollow noopener">medium.com/@madashindeina…

Make this and take my money.

PDF’s are dead. I’m having CC creating POCs, so I can a link for something people can play with.

@mattpocockuk Handoff should be super helpful for long grilling sessions 🔥

Two new skills this week! - /handoff compacts your current session to a markdown file - /prototype helps you prototype anything - UI or backend And a bunch more updates. Here's the changelog:

@mattpocockuk Microsoft's superpower has revealed itself once again. Btw talking about grilling, what you usually do, if in the middle of a grilling session you find agent hitting dumb zone already?

Microsoft Edge does not like /grill-me, apparently

Agent's smart / dumb zones @mattpocockuk talks about is an absolute reality, not paranoia. Claude code Opus 4.7 1m. On 150k tokens (15%), it ignores half of the statements in AGENTS.md and its references. >.<

A way to accidentally let your .env variables leak: - .env file opened with highlighted env value. - start Claude Code terminal discussion. 🫠 Mechanism: IDE selection -> system-reminder payload -> conversation history. Bypasses file read and tool restrictions.

4 of the most confusing terms in AI, defined: Model: a blob of parameters, written during training. Does next-token prediction and nothing else. Stateless. Harness: everything around the model that turns it into an agent: tools, system prompt, context window management, etc. Environment: the world the agent acts on. Anything outside the harness that the agent perceives and acts on via tools. Agent: a model, harnessed, in an environment. --- Opus is a model. Claude Code and Claude Web are different agents, because their harnesses differ - even though the models are the same. The file system is an environment. MCP servers add tools to the environment.

@mattpocockuk Is there any option to get info about updates of that skills? 🤔 There was a case when /grill-me was improved, but I was still using outdated version.

Nearly 23K stars for a collection of markdown files I wrote I guess they must be pretty good I want to invest more time in this repo. So, folks who starred it, what can I do to make these skills more obvious to you? - A docs site for the skills? - Send them to plugin marketplaces? Help me help you github.com/mattpocock/ski…

🚨 Vercel was hacked today. You may need to take some actions: check activity logs and rotate envs (or replace them with sensitive ones). Details there: lnkd.in/dKtPY-Rs #vercel #security #hacked

@mattpocockuk Seem to be nice in a combo with caveman =)

Anyone who's tried my new /domain-model skill - any feedback? Hearing positive noises but would love more detail. It replaces /grill-me in my stack, adding a thin layer of docs and ADRs during ideation.

My new skill lineup: /domain-model - replaces /grill-me, integrates some DDD concepts and adds docs & ADR's during discussions /to-prd - create a PRD /to-issues - create issues with blocking /github-triage - triage issues with a state machine-based labelling system /tdd - do TDD where appropriate Still more to flesh out, but this is feeling AWESOME

@mattpocockuk As a side effect, for the past couple of days, I’ve been answering the agent’s questions sleeping for eight hours straight. Sometimes I don’t have an answer, and my sleep falls into a cycle where I keep going back to the same thing. Fever dreams without fever. Not fun at all. 😐

@mattpocockuk My life changed after /grill-me. Planning now is so comprehensive! That’s the best way to make agent do really what I want to do. Having 8-10 sessions daily, sometimes in parallel. But the opposite side of that productivity - huge mental load. Taking decisions for 4h+ is hard.

'Improve the way you prompt the agent' I've started to consider this advice a bit of a red flag 'Commanding' the agent isn't the right approach. Instead, collaborate with it. Talk to it. Reach consensus together. /grill-me is great for this.

@mattpocockuk Well, indeed. My "global" version contains that and "workspace" one probably have an earlier version 🤔

@MadaShindeInai It does do that

I've found myself writing: "I don't know this area of code well. Go up a layer of abstraction. Give me a map of all the relevant modules and callers." Might need a new skill here. What should I name it?

The only #Mythos I want to hear news about. #claude #ai

@mattpocockuk Just presented Sandcastle to my collegues today 🙂 However, for me it do not work out of a box on my mac and require ~1h of try-fix-try session. Anyway - great tool! Thanks a lot!

I built a framework for co-ordinating AFK coding agents. It's called Sandcastle. Watch me use it to pick tasks, parallelize N coding agents, and merge the code - all AFK: