Fun fact - Even all the videos for my app are actually created using my app itself. Talk about eating your dog food. What do you think? Would love and appreciate an honest review. 🙏 https://www.youtube.com/@vois-so

@Md Jahid Hasan correct, but it doesn’t replace api use cases that may need real time generation

@Hicham Char How much are you able to get out of those credits, I tried and even for testing what it sounds like you have pay credits.

Love the excitement around Clawdbot - it really does feel like the "Jarvis moment" we've been waiting for. The potential here is genuinely exciting - this does feel like a turning point for AI in workflows. That said, I'd encourage caution before connecting it to production systems. There's been a wave of incidents this past week: Shodan scans showing 950+ exposed instances with open ports and no auth, brute force attacks targeting self-hosted servers, prompt injection via email causing data leaks, and reports of agents going rogue - mass-messaging customers, flooding social media with thousands of comments. Daniel Miessler just published a top 10 security hardening guide that's worth reading before deploying. I've been building a similar GTM automation engine, and these reports validate the architectural choices I made early on: queue-based publishing with mandatory human approval gates, no direct "send now" capability, and hard blocks (not just warnings) on outbound actions. The agent drafts content, but nothing leaves without explicit human sign-off. It adds friction, but friction is the point when the alternative is an agent autonomously sending thousands of messages. The tech is powerful. We just need to treat it like the sharp tool it is. So I ran a deep code analysis on my GTM engine code base against Clawdbot's codebase and found these vulnerabilities. I thought I’d share a word of caution. I know people are going crazy about Claude Bot, but make sure you know exactly what you’re doing.