German-language AI community and magazine, with a particular focus on AI security & pentesting — a niche with very little German-language coverage.
- 🌐 Website & blog: flowki-club.de
- 💬 Discord ("Hacking & Security" zone and others): flowki-club.de/join
Real, working code and configs extracted from published FlowKI Club articles — not demos. Each repo faithfully reflects what's actually shown in its linked article; where an article's code was incomplete, we said so instead of inventing the rest.
| Repo | What it is |
|---|---|
| flowki-bugbounty-recon | LLM-orchestrated bug bounty recon pipeline (amass/subfinder/httpx/nuclei) + OSINT extension |
| flowki-llm-redteam-checklist | 50-point LLM red-team checklist (Input/Auth/Output/Tool-Use/RAG/Cost-DoS) with severity ratings, plus an honestly-labeled starter runner |
| flowki-llm-safety-pipeline | Complete 6-stage LLM safety middleware (rate-limit, sanitization, injection detection, prompt hardening, output validation, audit log) — Python + TypeScript, benchmarked against PurpleLlama/AdvBench |
| flowki-claude-code-hardening | Deny-list config, 4 documented attack patterns, and a pentester checklist from a hands-on security review of Claude Code's permission architecture |
| flowki-stalkerware-scan-toolkit | MVT + Claude Code wrapper scripts to check iOS/Android for stalkerware, with warning signs, prevention, and crisis-support resources |
| flowki-knowledge-mcp | MCP server exposing the FlowKI Club blog as a live knowledge base (list/search/read articles, list pillars) — queries flowki-club.de directly, no content mirror |
Eight content pillars, all in German: KI-Pentesting & Security, Coding mit KI, Business mit KI, KI-Einstieg, KI-Modelle & Benchmarks, Crypto mit KI, Content mit KI, B2B-Lead-Generation.