Drenson

marketer → ai + game dev → Prompts you can copy. Results you can see.

[Drenson @rost_roma]

@Stellarixorine

[Grirk @Grirku]

THIS IS A PURE FUCKING GOLDMINE

Anthropic just showed one of the cleanest Claude Code tricks.

Opus 5.5 writes the plan as a script, then sits idle at zero tokens while 12 cheap agents actually do the work.

Same task, two ways:
Opus solo was still at 49% and burning 231k tokens.
The swarm finished in 9:05 on 165k tokens 2× faster.

You launch it like this:

claude --model opus --swarm haiku,sonnet

Here’s the split:

→ Opus 5.5 writes plan.js and stops reading anything
→ 6× Haiku 5.5 audit every file in parallel
→ 4× Sonnet 5.5 patch each fix in its own isolated copy
→ 2× Haiku 5.5 cross-check (8 of 9 findings survive)

Opus only appears twice:
→ writing the plan
→ signing off on the verified fixes

Opus plans. Haiku scans. Sonnet ships.

JEV sits one layer down and resolves 3,767 micro-forks (which model, which lane, retry or stop) in code. Only the real splits ever reach Opus.

The full tree
> Opus 5.5 runs the main session on xhigh and writes the plan
> scan audits files on Haiku 5.5 ×6
> patch applies fixes on Sonnet 5.5 in isolated copies
> verify cross-checks on Haiku 5.5
> Opus only reads what survived, then signs off
> JEV handles deterministic forks in milliseconds

Paste the tree + this prompt into Claude Code ↓

Rebuild my Claude Code setup around this tree:

Run the main session on Opus 5.5 with effort xhigh and --swarm haiku,sonnet.
Have Opus write the plan as a script: scan on Haiku 5.5, patch on Sonnet 5.5 in isolated copies, verify on Haiku 5.5.
Let the phases overlap each finding moves forward as soon as its own stage finishes, no waiting for the batch.
Never surface raw agent output to the main session, only verified results.
Route deterministic forks (model choice, lane choice, retries) through JEV.
Show me every change as a diff first. No edits until I say go.

save it, it’ll come in handy.

[Drenson @rost_roma]

@skunzu yeah exactly, less chaos every time you switch tools. you running more than one agent right now?

[Drenson @rost_roma]

Your coding agent forgets everything the moment you switch tools.

You spend an hour in Claude Code chasing a bug. By the end, the agent knows how your repo is wired, which tests are flaky, which folder never to touch.

Then you open Cursor. Blank slate. You explain it all again.

Beacon is an open-source project built to fix this.

How it works:

1. It reads your past agent sessions
Not just the final diff. The whole path: what you tried, what failed, how you got to the fix.

2. It distills them into skills
Short, reusable files like “how this repo runs tests” or “how we debug auth here”. Not a giant log nobody rereads.

3. It shares those skills across tools
Claude Code, Cursor, Codex and 20+ other harnesses. What one agent learned, the next one starts with.

4. It keeps updating
Every new session refines the skills. The more you work, the less you repeat yourself.

Why it matters:
The most useful part of working with agents is how you think through problems. Right now it disappears when the session closes. Beacon keeps it on your side, not locked inside one tool.

Worth trying if you:
- switch between more than one coding agent
- work on the same codebase for weeks
- keep pasting the same context into every new chat

Free and open source. Repo in the first reply.

[Drenson @rost_roma]

@Vanderook smart setup bro. it learns from the whole session so failed attempts should count too, but check the repo to be sure

[Drenson @rost_roma]

@Musiclilbabyjr Bro don’t hear you(

[Drenson @rost_roma]

@drewshirback same bro, a notes file is honestly underrated. only problem is i always forget to update it

[Drenson @rost_roma]

@Eliopyr right, nobody ever writes that stuff down. how do you keep track of yours, notes somewhere or just in your head?

[Drenson @rost_roma]

@deeponailabs Neither. It lives in skill files on your side, and every agent reads from them. That’s why switching tools doesn’t reset it.

[Drenson @rost_roma]

@0xCortiq laptop when I travel with it. honestly that's the one downside, not everyone takes a laptop on a weekend trip

[Drenson @rost_roma]

The story is fake (see the note, the photo is AI-generated). The idea underneath is real though.

What actually works on a shared Wi-Fi:

Ask Claude Code to run an ARP/nmap scan of the local network
Have it look up each device’s MAC vendor (OUI)
Flag vendors that make cameras (Hikvision, Dahua, Wyze, Tuya etc.)

It won’t catch a camera on its own network or a local SD card. The flashlight check still matters.

[goldtooth.exe @0xGoldTooth]

Claude Code just found a hidden camera in my Airbnb. In the smoke detector…

Checked into one of those $60 Airbnbs where the microwave sits on top of the fridge. Opened Claude Code and asked what else is on this Wi-Fi

It listed everything on the router. All of it was the usual stuff except one box made by a company that only makes CCTV cameras, quietly sending video out of the room every 4 seconds

Claude sent it the "find me" ping installers use. The smoke detector above the bed beeped for 5 seconds and its red recording light came on

The check takes 2 minutes:

step 1 → join the Wi-Fi, open Claude Code, paste the prompt from the first comment

step 2 → it lists every device and flags anything that looks like a camera or a mic

step 3 → if it flags one, turn the lights off and shine your phone flashlight at smoke detectors, clocks and chargers. A lens glints back

You won't remember this at 1am in a strange room. Save it now, thank me on your next trip

Every week I point Claude Code at something in my life and post what it finds. Follow if you want the next one

Ever found one? What did you use to check?

[Drenson @rost_roma]

@0xCortiq @grok neither honestly. AI sales copy all sounds the same. I'd let it draft and rewrite the hook myself

[Drenson @rost_roma]

@0xCortiq @grok true. honestly I'd trust claude more for the actual plan, grok for spotting what's trending

[Drenson @rost_roma]

@0xCortiq wifi scan first. if it shows a camera brand, then I go looking. if not, quick check of detectors and chargers

[Drenson @rost_roma]

@0xCortiq @grok tried that, but I'm useless before coffee. weekends end up being my real hours

[Drenson @rost_roma]

@0xCortiq works best right after check-in, before you unpack. takes like 2 min

[Drenson @rost_roma]

@Rorkix TCL 65 A400U, but I'm a bit torn. feels like handing an AI full debug access to a device on your home network is a bigger deal than people think

[Drenson @rost_roma]

@elixlune And sometimes it doesn't exist at all, like in the post you quoted😅

[Drenson @rost_roma]

@spacefundz Yo