Bol som zrusit ucet v nemenovanej obchodnej banke. Pani za prepazkov ma presviedcala aby som si nechal u nich ucet. Dala mi pritom 10 minutovy monolog v ktorom som sa dozvedel, ze som blbec, ze rusim u nich ucet, ktory urcite budem potrebovat, ze ine banky, ktore mam nie su normalne banky, a pod. No normalne mi az za nu trapne bolo co to na mna skusa. Akoze kto ich toto skoli? Je to grc
pre starsie rocniky vidim ze bola posilnena MHD ale tento rok nevidno ani prd. Bude nejako posilnena?
Chcem potahat klimu po viacero izbach ale musel by som robit drzaky do zelezobetonu, do plafonu, hladat kade ide elektrika k lampam a pod... Ako ste to vsetko riesili?
Ako hlboko ste robili drazky? Ako daleko moze byt vnutona jednotka od vonkajsej?
Kludne mimo SK... Kde najblizsie?
Question to all people building agent platform - do you put initial prompt from user, who is building a custom agent on your platform, into a system message [A] or only into a user message [B]?
If you put it into user message - how do you hide it in UI?
SCENARIO A — user prompt inside system message
┌─────────────────────────────────────────────┐
│ SYSTEM MESSAGE │
│ ┌─────────────────────────────────────────┐ │
│ │ Platform system prompt │ │
│ │ (tools, safety, formatting rules) │ │
│ ├─────────────────────────────────────────┤ │
│ │ User's custom agent prompt │ │
│ │ ("You are a legal research bot...") │ │
│ └─────────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
┌─────────────────────────────────────────────┐
│ USER MESSAGE 1 │
│ "Summarize this contract." │
└─────────────────────────────────────────────┘
│
▼
┌───────────────────────┐
│ MODEL │
└───────────────────────┘
SCENARIO B — user prompt in first user message
┌─────────────────────────────────────────────┐
│ SYSTEM MESSAGE │
│ ┌─────────────────────────────────────────┐ │
│ │ Platform system prompt │ │
│ │ (tools, safety, formatting rules) │ │
│ └─────────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
┌─────────────────────────────────────────────┐
│ USER MESSAGE 1 │
│ ┌─────────────────────────────────────────┐ │
│ │ User's custom agent prompt │ │
│ │ ("You are a legal research bot...") │ │
│ ├─────────────────────────────────────────┤ │
│ │ Actual request │ │
│ │ "Summarize this contract." │ │
│ └─────────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
│
▼
┌───────────────────────┐
│ MODEL │
└───────────────────────┘
Question to all people building agent platform - do you put initial prompt from user, who is building a custom agent on your platform, into a system message [A] or only into a user message [B]?
If you put it into user message - how do you hide it in UI?
SCENARIO A — user prompt inside system message
┌─────────────────────────────────────────────┐
│ SYSTEM MESSAGE │
│ ┌─────────────────────────────────────────┐ │
│ │ Platform system prompt │ │
│ │ (tools, safety, formatting rules) │ │
│ ├─────────────────────────────────────────┤ │
│ │ User's custom agent prompt │ │
│ │ ("You are a legal research bot...") │ │
│ └─────────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
┌─────────────────────────────────────────────┐
│ USER MESSAGE 1 │
│ "Summarize this contract." │
└─────────────────────────────────────────────┘
│
▼
┌───────────────────────┐
│ MODEL │
└───────────────────────┘
SCENARIO B — user prompt in first user message
┌─────────────────────────────────────────────┐
│ SYSTEM MESSAGE │
│ ┌─────────────────────────────────────────┐ │
│ │ Platform system prompt │ │
│ │ (tools, safety, formatting rules) │ │
│ └─────────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
┌─────────────────────────────────────────────┐
│ USER MESSAGE 1 │
│ ┌─────────────────────────────────────────┐ │
│ │ User's custom agent prompt │ │
│ │ ("You are a legal research bot...") │ │
│ ├─────────────────────────────────────────┤ │
│ │ Actual request │ │
│ │ "Summarize this contract." │ │
│ └─────────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
│
▼
┌───────────────────────┐
│ MODEL │
└───────────────────────┘
Kupil som stary zhnity dom aj s pozemkom od statneho podniku v ktorej je podnajomnik strasne dlho. Najomna zmluva ma ~25 rokov a je podla zak. c. 18/1996 Z.z., predpokladam, ze to boli este nejake kvazi najomne byty alebo co. Najomnik je strasny magor, nechce ma tam pustit, nechce upravit zmluvu, ani nic... a podla povodneho majitela sa ho nemohli zbavit. Teraz ked to preslo na mna ako FO ako sa ho mozem zbavit?
I have configured it to be english but everything is still in german. Why? How to fix it?
(company is in germany, I have my private PC in english, I don't have germany configured anywhere)
Majitelia ESET, Penta, etc... investuju prachy do zacinajucich startupov? Nejak nevidim takyto ekosystem u nas. Ci len sa o tom nepise? Su nejake startupy do ktorych investovali?
I use Claude Code in VS Code. These changes are really annoying. How to avoid it? Is VS Code auto format run on save? I have VS Code settings.json configured like this but I am not sure auto format is always applied and Claude is finished.
"[typescript]": {
"editor.tabSize": 2,
"editor.detectIndentation": false,
"editor.insertSpaces": true,
"editor.formatOnSave": true,
"editor.codeActionsOnSave": {
"source.organizeImports": "explicit"
}
},
"[typescriptreact]": {
"editor.tabSize": 2,
"editor.detectIndentation": false,
"editor.insertSpaces": true,
"editor.formatOnSave": true,
"editor.codeActionsOnSave": {
"source.organizeImports": "explicit"
}
},
All code in signup-form.tsx was generated from the beginning with the same VS Code formatting settings.
Typical agentic scenario - I need to create thousands or millions of databases for agents and users. Likely 99% would be very small and seldomly used - need to keep cost near 0. Remaining 1% will be traffic intensive - should not cost a fortune and likely should be capped somehow.
I was looking at Neon DB Agent Plan or Turso DBs - that look good. But is there anything in GCP that can be used here? I have no preference for DB type yet.
I am reading that I should design my own loops because that is the future. But what is it actually? Tell me about your useful loops you actually use. Or did you create any for someone else that they are using?