New: OMEGA Planning Mode
The most intelligent entry point on Linux. AI builds a 9-phase attack plan before running a single tool.
- cybermind /plan <target> — AI-first attack planning.
- 9 phases: OSINT → Subdomain → Ports → HTTP → Dirs → Vulns → Hunt → Secrets → Exploit.
- WAF-aware: Cloudflare, Akamai, Imperva bypass strategies built in.
- Target-specific: WordPress, GraphQL, JWT, AD, Cloud all handled differently.
- Auto-doctor runs before execution — missing tools installed automatically.
- System resource check before starting.
New: GitHub Models + Cloudflare Workers AI
Two new completely free AI providers added. More providers = more reliability.
- GitHub Models: GPT-4o, Llama 3.3 70B, DeepSeek R1, Phi-4 — 150 req/day free.
- Cloudflare Workers AI: Llama 70B, DeepSeek R1, Qwen 72B — 10,000 neurons/day free.
- Total: 11 providers, 50+ models in the fallback chain.
- Sequential fallback: Groq → Cerebras → OpenRouter → GitHub → Mistral → Cloudflare → ...
Fix: Cold start auto-wake
CLI now automatically wakes the backend and retries — no more manual resend.
- Shows live progress: ⟳ Backend waking up... (3s)
- Auto-retries after wake — user never needs to resend manually.
- Works on Windows, Linux, and macOS.
Security audit applied
Full security audit completed. Critical fixes applied.
- SSRF protection on all URL fetching.
- AI output sanitization — strips leaked secrets from LLM responses.
- Blocked device enforcement — 403 instead of log-only.
- Exponential backoff on API key brute force.
- GDPR data export endpoint added.
- SQL ambiguous column bug fixed.
What to do now
Update your CLI binary and try planning mode.
# Update CLI curl -sL https://cybermindcli1.vercel.app/install.sh | bash # Try planning mode cybermind /plan example.com