Next challenge at #BashLab: build an AI assistant for creating live OS with #livebuild.

I already built one for #Buildroot and one for #Pentesting.

Problem: no solid PDFs for live-build, only the official HTML docs with tons of links.
Now I need to feed the entire site into my RAG.

Time to scrape, structure, and index 🤟🏻

#AI #RAG #Linux #Debian #BashCore

BashLab just built a local AI assistant powered by RAG, trained on the best pentesting books. Thousands of pages indexed and ready 📚⚡

Paired with BashCore, this becomes an atomic combo 💣

Minimal OS + specialized AI = pure offensive & defensive power.

Independent. Local. Focused.

#BashLab #BashCore #CyberSecurity #Pentesting #RAG #AI #OpenSource

Parrot 7.1 is now rolling out 🥁

This release brings:

✔ MCPwn for LLM security tools execution
✔ New Community Spins: MATE, LXQt, Enlightenment
✔ Rocket 1.5.0 with a brand-new UI
✔ Bug fixes + smoother daily experience

Available now on the download page 🦜

Click the link and read more on our latest article 🔗

https://parrotsec.org/blog/2026-02-11-parrot-7.1-release-notes/

#ParrotSec #ParrotOS #linux #linuxdistro #cybersecurity #cybersecuritynews #debian #pentest #pentesting #hacking #hacker

New Tool Alert

pirebok - a guided adversarial fuzzer with evolutionary search

Give it a payload. It mutates it until it bypasses the WAF.

No wordlists. No manual tuning. 9 mutation strategies. Evolutionary search breeds payloads until they slip past classifiers.

pirebok -f GuidedRandomSqlFuzzer -p "admin' OR 1=1#"
> aDm1n'/**/oR/**/0x1=1#

pip install pirebok
github.com/HappyHackingSpace/pirebok

#infosec #waf #sqli #fuzzing #pentesting #python #opensource #HappyHackingSpace #Diyarbakir

This security researcher just released v1.1.0 of a fully automated pentesting platform dubbed "RedAmon". Manage 6 stage reconnaissance, AI agent metasploit attacks and launch attacks from a web UI that lets you manage the tests as projects.

I haven't dug in yet but looks pretty impressive.

#cybersecurity #pentesting #CISO

https://github.com/samugit83/redamon

Prompt Worms Часть 2: Я проверил на практике — 31 уязвимость в экосистеме AI-агента

В первой части мы разобрали теорию Prompt Worms — самореплицирующихся атак через AI-агентов. OpenClaw был назван «идеальным носителем». В этой части я проверил на практике: скачал репозиторий, залез в код, прощупал инфраструктуру и нашёл 31 уязвимость в 4 слоях экосистемы. Ноль санитизации на 867 строк мозга, timeout вместо approval, бэкдор-«пасхалка» в коде, утечка хешей паролей в маркетплейсном SaaS, и 14 слепых зон в их собственной threat model. Три дня, ~4,500 строк трассировки, 14 kill chains с PoC

https://habr.com/ru/articles/994230/

#prompt_injection #ai_agents #llmattack #openclaw #ai_security #threat_model #supply_chain_attack #mitre_atlas #pentesting #prompt_worms