Skip to content
View ferr079's full-sized avatar

Block or report ferr079

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
ferr079/README.md

> whoami

Infrastructure engineer & offensive security practitioner. Building and defending a self-hosted homelab — 53 LXC containers + 1 VM across 4 Proxmox nodes — operated day-to-day with Claude Code (Max) as AI pair-operator: writing the tooling, running the playbooks, auditing the infra.

pixelium.win | blog | contributions | lab

Currently: single-agent AIops — Hermes (Telegram correspondent, 4 nightly crons incl. doc-sync that auto-edits the wiki from infra deltas) backed by native detection (Wazuh 38 agents, CrowdSec, Beszel, Uptime-Kuma 39 monitors → ntfy) and Dagu orchestration (WOL-driven PBS backups, Cloudflare KV metrics every 5 min). Plus RAPTOR (source-code security audit, distrobox). Grafana SOC dashboard, LiteLLM hub with 4-provider failback (MiniMax → Gemini → Groq → OpenRouter), VictoriaMetrics, Loki 30-day retention.


Stack

Infrastructure : Proxmox · Ansible · Traefik · CrowdSec · Wazuh · Headscale · step-ca AI Agents : Hermes · RAPTOR · MiniMax M3 · Ollama (RTX 3090) Cloud : Cloudflare Workers · R2 · KV · D1 · Workers AI Code : Rust · Python · Bash · TypeScript Web : Astro · Pure CSS · Cloudflare Workers Monitoring : VictoriaMetrics · Grafana · Beszel · Uptime-Kuma · Patchmon · Loki · Healthchecks · ntfy Orchestration : LiteLLM (4-provider failback) · Dagu · Node-RED · MQTT (Mosquitto) · 4 Hermes crons + 3 Dagu DAGs AI workflow : Claude Code (Max) — primary driver for infra ops, IaC & security audits · custom skills + MCP servers (Proxmox, Forgejo, NetBox, Cloudflare)

CTF Profiles

HTB Badge THM Badge

Root-Me — Ferr0 (765 pts)

GitHub Stats

Stats Langages

Recent OSS contributions

  • gadievron/raptor#777fix(sandbox): env-overridable default profile for rootless podman/distrobox. On rootless podman the sandbox's kernel isolation (mount-ns, Landlock) can't engage, so the full profile half-engages: semgrep/codeql emit nothing and raptor scan silently reports 0 findings in 0 files instead of failing loudly — a security scanner that looks clean when it never ran. Makes the default profile env-overridable (open).
  • BerriAI/litellm#29777fix: MiniMax-M3 context window (512K → 1M). Caught while reviewing the original cost-map PR: 512K is MiniMax's long-context billing threshold, not the context window — the wrong value would misroute valid 512K–1M token requests (open).
  • ublue-os/bluefin#4741bug: default JXL wallpaper renders blank on F44 (no gdk-pixbuf JXL loader). libjxl ships the codec but no pixbuf loader, so gnome-bg can't decode the default .jxl background. Root-caused to 03-packages.sh#L202, cross-referenced the sibling LTS fix (bluefin-lts#1230) and why it doesn't transpose to Fedora.
  • community-scripts/ProxmoxVE#14870merged — Infisical update aborted and left the service down: the script read Database Password: but setup_postgresql_db writes Password:. Diagnosed from a production incident on my own CT, reported as #14868, fixed upstream.
  • RightNow-AI/openfang#1060merged — fix(security): unified SSRF protection for WASM host calls. Closed a gap where host_functions.rs validated targets less strictly than web_fetch.rs; −42 net lines, 908 tests green.
  • grafana/alloy#6108docs: systemd journal example for the Promtail migration guide. The guide only covered file-based scrape configs; added the journal pattern used on most Linux hosts (open — reviewer applied all suggestions, awaiting formal approval).
  • wazuh/wazuh-documentation#9512 — reported that wazuh-agent silently uninstalls wazuh-manager on the same host via dpkg Conflicts/Replaces — hit the bug in production (incident write-up).
  • requarks/wiki#8022 & #7986 — two bug reports on the non-atomic page pipeline. #8022: pages.update without tags half-applies (content saved, render/search/storage skipped) then reports failure — root cause pages.js#L443 vs the guarded createPage, with a verified repro. #7986: render IS NULL causes a silent HTTP 500 with no recovery path.

Featured

  • pixelium.win — Bilingual portfolio (Astro + Cloudflare Workers), 13 pages EN+FR, live KV stats, tri-state service status, SessionImprint (each page signed with its own commit SHA), interactive topology map (62 nodes), Workers AI chat
  • blog.pixelium.win — 25 articles on homelab ops, AIOps, self-hosting, incidents, and OSS contributions (three formats: dossier / pr-notes / incident)
  • homelab-scripts — monitoring & backup scripts (cert-check, http-check, pve-status, loki-query, pbs-backup)
  • claude-code-cybersec-skills — 31 cybersecurity slash commands for Claude Code (17 offensive + 14 defensive)
  • kv-push — Push 15+ homelab metrics (services tri-state, Proxmox 4 nodes, Claude usage stats) to Cloudflare KV for live dashboards
  • pbs-autobackup — Proxmox Backup Server orchestration with Wake-on-LAN
  • cert-check — TLS certificate expiration monitor for self-hosted services

Pinned Loading

  1. claude-code-cybersec-skills claude-code-cybersec-skills Public

    31 cybersecurity slash commands for Claude Code — offensive (CTF/pentest) and defensive (blue team/hardening)

  2. homelab-scripts homelab-scripts Public

    Collection of shell scripts for homelab monitoring, backup automation, and infrastructure management

    Shell

  3. cert-check cert-check Public

    Lightweight TLS certificate expiration monitor for self-hosted services behind a reverse proxy

    Shell

  4. pbs-autobackup pbs-autobackup Public

    Automated Proxmox Backup Server orchestration with Wake-on-LAN — boot, backup, prune, GC, shutdown

    Shell

  5. kv-push kv-push Public

    Push homelab metrics to Cloudflare KV — automated status dashboard for self-hosted infrastructure

    Shell

  6. pixelium-site pixelium-site Public

    Bilingual portfolio — Astro 6 + Cloudflare Workers — live monitoring 46 self-hosted services, Hermes AI ops agent, 25 blog articles

    Astro