Candle Code retweeted
Claude now has a line item for “personal or off-topic.” Here is what’s in it🤨
Your admin can now see the bill for your off-topic Claude chats. Not the chats.
Claude Enterprise “smart reports” (generally available since Oct 28) include an Inefficiencies section: sessions that produced nothing usable, and sessions that were personal or off-topic.
What it shows: a count and a cost at list price.
What it does not: summaries or drilldown. Names show as “User” unless the company turns attribution on, and every reveal is logged. Reports are off by default.
A sample screenshot shows 18 off-topic sessions costing $217 vs 115 failed sessions costing $106. My math: about $12 per off-topic session vs $0.92 per failed one. (Screenshot source not confirmed.)
My take: a cost counter, not a spy cam. Anthropic says reports should not be used for performance reviews. A tech outlet notes nothing technical enforces that.
Is this a fair way to track AI spend, or too much oversight?
Eu sempre gostei de programar, mas a IA tá mexendo com isso. Você passa anos aprendendo lógica, banco de dados, resolvendo bug de madrugada, aí alguém que nunca escreveu uma linha abre o Claude e faz algo parecido numa tarde. Eu sei que não é a mesma coisa, mas dá uma desanimada.
Candle Code retweeted
Working with agents all day is more tiring than writing code ever was, and we must admit it.
Typing wasn't that tiring, you knew what you wanted, your hands did it, your head and thoughts wandered a bit.
My days now are 3-4 agents running all the time, and each one stops every few minutes with a question. Approve this command? Option A or B? This test fails, should I change the test? Should I continuue?
Each one is a an actual small decision, but each one needs the full context of a codebase and new code I didn't write line by line.
By 6PM I'm done, not because I wrote a lot, but because I decided a lot. For a while I thought my limit was tokens, then the weekly usage limit, but turns out it wasn't. My limit is how many good decisions I can make before I start approving everything just to make the questions stop.
How many agents do you run in parallel and for how long before your own answers get worse?
Candle Code retweeted
STOP opening links on X until you see what this one knows about you
your city. your internet provider. your phone. that you came here from X. on android even your battery %
you don't type anything. you don't allow anything
works on iPhone, iPad, android, windows, mac and linux. every one of them hands this over the second a page loads, just different pieces
and incognito won't help. your city and provider show up exactly the same, and the device ID it builds usually stays the SAME too. that's how ads keep following you after you clear cookies
i built it so you can see what every site sees. nothing is saved, it all runs in your browser
there are 4 ways to show less at the bottom. a VPN alone kills the scariest part
open it and tell me what it got wrong
link in the first reply. bookmark it before you send it to the group chat
Candle Code retweeted
Nobody wants to say this out loud.
$BTC just dropped from $81K to $75K. Down 6% in a single day.
Timeline is full of "it's over" posts. Same energy as every other dip.
Every crash feels like the end while you are inside it. Every single one.
Save this post. Read it again in 6 months.
Candle Code retweeted
Your self-worth becomes fragile when you outsource its measurement to people who barely know you.
Candle Code retweeted
THIS ROBOT IS PICKING ORDERS FASTER THAN YOUR WAREHOUSE GUY ON HIS THIRD COFFEE
Dark warehouse, red shelving, blue bins everywhere, and a white two-armed robot rolling down the aisle like it owns the place. It pulls bottles and boxes off the shelves one by one, drops them into the carton and moves on without a single wasted motion. No humanoid face, no fancy walking, just two arms on wheels doing the most boring job in logistics. That's exactly why it's scary good, the boring jobs are the ones that actually get automated first. Nobody's gonna pay for a robot that does backflips, they'll pay for one that picks 1000 items a shift and never calls in sick
Candle Code retweeted
⚡ Autonomous AI agents just had their npm moment
Perplexity Computer rolled out Agent Plugins, allowing users to import entire modular skill and connector bundles directly from GitHub or zip packages
Instead of developers manually configuring twenty API endpoints, writing custom tool definitions, and tuning system prompts for every workflow, agent capabilities are now packaged as single portable units
A single plugin bundles your third party authentication tokens, private enterprise connectors, and execution procedures into one drag and drop package that works across agent harnesses
We are officially transitioning from prompt engineering to package management for digital workers
Instead of hiring or hand crafting an agent from scratch, teams will simply clone open source plugin repositories to deploy full stack researchers, code reviewers, or growth operators in sixty seconds
The real bottleneck is no longer building agent harnesses, it is securing arbitrary community plugins from supply chain prompt injections
Are we ready for open source package managers where dependencies have autonomous shell and browser access?
Polymarket Event Info Bot: Instant Edge for Traders
t.me/M1PolyRadar_bot
Polymarket TG bot: data in 2s. Paste link → instant summary. No browser.
Polymarket Bot Features
•Instant Yes/No probs, volumes, liquidity, top 5 markets
•Total volume, liquidity ratios, close dates
•Smart cache: instant repeats (5min)
•Inline buttons: View/Delete
Hooks traders: fast price checks, sentiment scans, cross-market comps
Free, private, unlimited
Level up Polymarket with M1!






















