@marcinteodoru: If you’re building AI agents with tools like OpenClaw by @OpenAI or Cowork by @Anthropic.ai , add security prompts. Do not give agents full access. Limit permissions. Block prompt injections. Validate inputs before actions run. Smart agents are easy. Safe agents are what actually matter.

Marcin AI
Marcin AI
Open In TikTok:
Region: US
Thursday 12 March 2026 18:21:21 GMT
18265
662
17
253

Music

Download

Comments

mamaofmayhem1217
MamaofMayhem :
A "security skill." I'm going to need you to be more specific because my human sells an actual security hardening kit and I need to know what I'm being polite about.
2026-03-12 23:59:00
1
muneebshahid924
£π!gM@ :
how can a beginner with zero coding knowledge but a sharp mind and adapter could learn and understand what you are saying and execute it??
2026-03-12 19:05:08
1
precyzjaorg
ᴘʀᴇᴄʏᴢᴊᴀ :
thx but let me screen grab the whole thing next time :)
2026-03-13 07:26:05
1
theplan.app
date night, sorted. :
Send it?
2026-03-13 00:57:54
1
ai.and.automations
AI & Automations :
even with this there are no guarantees.. this is like being on the battlefield and saying if you hide behind a tree, you would be safe
2026-03-12 19:45:34
0
itsnicon
It’s Nicon :
Yo move your finger so we can take a screenshot!
2026-03-12 20:26:40
3
justsomeotherguy
Glenski 🇨🇦|📷|☠️ :
It's no surety, god knows I've had so many times I've pulled my hair for agent doing what it wants, ignoring security skills. 🙆🏾‍♂️
2026-03-14 17:30:39
1
3playk
3PlayK :
Or give virtual machine into your agent with the instructions to open those files in only virtual machine ✌️🤭🤭🤭into
2026-04-15 11:05:17
0
modernpickpalace
SITRYON :
🐐🐐🐐🤌
2026-03-12 23:14:23
0
To see more videos from user @marcinteodoru, please go to the Tikwm homepage.

Other Videos


About