root@rebcesp
3.2K posts

root@rebcesp รีทวีตแล้ว

Royal Match is arguably the most successful mobile app ever created by a turkish company
$67,000,000/month
Insane

Jacob Rodri@jacobrodri_
Turkey in mobile gaming is insane: 2010: 3 turkish founders start an app studio called Peak Games 2020: They sell it to Zynga for $1.8B 2019: before the exit, Peak employees start another app studio (Dream Games) 2026: Dream Games is now valued at $4B They're built different
English
root@rebcesp รีทวีตแล้ว

@damian_89_ @thedawgyg Does RavenX-CyberAgent-v6.2 run properly on an RTX 3090 (24GB VRAM)?
English

@thedawgyg ravenx-cyberagent-v6.2,qwen3.6-27b-neo-code,gemma-4-26b-uncensored - for bug bounty hunting - extremly good, usually better then deepseek.
English

@4osp3l Sometimes it can be useful to leverage Google-owned domains, such as Google Cloud Shell, and keep a server running there, ready to receive any requests or data you need.
English

@lonelysloth_sec to run models at home? your best bet is 2x 5090 at this price range I think, but you're just below the b200 price. then a good enough storage for the 1T+ models. keep in mind parameters are up only, sooner or later you'll have to run 5T+, maybe 10T+
English
root@rebcesp รีทวีตแล้ว
root@rebcesp รีทวีตแล้ว

Claude code’s /security-review is just a Skill, and the whole prompt is in this repo
It’s p generic and imo you can tailor it to each repo to language you’re scanning to get better results
github.com/anthropics/cla…
English

Do you watch Netflix in your free time?
Try hackflix for security conference talks
h4ckfl1x.com
#cybersecurity #bugbounty

English
root@rebcesp รีทวีตแล้ว

root@rebcesp รีทวีตแล้ว

A Claude Code skill bundle for bug hunting and external red-team work - 51 skills, 15 slash commands, 574+ disclosed-report patterns curated across 24 vulnerability classes, plus enterprise identity + infrastructure attack matrices. github.com/elementalsouls…
English
root@rebcesp รีทวีตแล้ว


Criei uma ferramenta que:
- Coleta rotas, APIs, JS, DOM, screenshots, requests e responses
- Recebe diferentes usuários e compara o comportamento entre eles
- usa AI pra navegar pela aplicação interagindo como um usuário real
- Entende telas, fluxos, ações disponíveis e contexto real
- Guarda todo o contexto obtido em grafos de evidências
- Faz múltiplos agentes debaterem por turnos sobre como prosseguir
- Obriga cada agente a defender hipóteses com evidências reais. Eles tem 6 turnos pra convencer os outros que a ideia vale a pena. Cada um tem uma ""personalidade"" diferente
- Manda as hipóteses aprovadas pra agentes especialistas por classes de vulnerabilidade
- Usa skills de +10k linhas cada pra criar payloads, bypasses e validações de acordo com a aplicação
- Tenta explorar a vuln
- Se conseguir, volta pro debate pra pensar em possíveis chains com outras vulns
Até agora tá trazendo resultados legais!! quem sabe nao deixo publico essa brincadeira
Português














