# SecuBox UI & Theme History

_Last updated: 2026-02-07_

1. **Unified Dashboard Refresh (2025-12-20)**  
   - Dashboard received the "sh-page-header" layout, hero stats, and SecuNav top tabs.  
   - Introduced shared `secubox/common.css` design tokens.

2. **Modules & Monitoring Modernization (2025-12-24)**  
   - Modules view adopted the same header/tabs plus live chip counters.  
   - Monitoring cards switched to SVG sparkline charts with auto-refresh.

3. **Alerts + Settings Overhaul (2025-12-27)**  
   - Alerts page now mirrors the dashboard style, dynamic header chips, and filtering controls.  
   - Settings view gained the SecuNav tabs, chips, and shared design language.

4. **Theme Synchronisation & Deployment (2025-12-28)**  
   - All SecuBox views call `Theme.init()` to respect dark/light/system preferences.  
   - Navigation bar now darkens automatically for dark/cyberpunk themes.  
   - Monitoring menu entry simplified (no `/overview` shim) to prevent LuCI tab duplication.

5. **Router Deployment Notes**  
   - Use `secubox-tools/deploy-secubox-dashboard.sh` for view-only pushes.  
   - Use `secubox-tools/deploy-secubox-v0.1.2.sh` for RPCD/config updates.  
   - Always clear `/tmp/luci-*` after copying UI assets.

6. **SecuBox v0.5.0-A Polish (2025-12-29)**
   - Monitoring and Modules views drop legacy hero/filter UIs; all tabs now use SecuNav styling.
   - Help/Bonus page adopts the shared header, navbar entry, and chips.
   - Alerts buttons use `sh-btn` components; nav + title chips inherit theme colors.

7. **Multi-Instance Support (2026-01-20)**
   - CrowdSec LAPI port configuration fix for multi-instance deployments.
   - Streamlit and HexoJS gain multi-instance management support.
   - HAProxy enhanced with instance-specific configuration.

8. **HexoJS Build & Publish Integration (2026-01-21)**
   - Added LuCI interface for Gitea-based Hexo build and publish workflows.
   - Automated Git operations for static site generation.

9. **ARM64 Toolchain Build Requirement (2026-01-27)**
   - Discovered SIGILL crashes on ARM64 (MochaBin) due to LSE atomics in SDK-built Go binaries.
   - Documented requirement: Go/CGO packages (crowdsec, netifyd) MUST use full OpenWrt toolchain.
   - SDK produces binaries with LSE atomic instructions that crash on some Cortex-A72 CPUs.
   - Updated CLAUDE.md, secubox-tools/README.md with toolchain build rules.

10. **Documentation Regeneration (2026-01-27)**
    - README.md updated to v0.16.0 with all 38 modules categorized.
    - Added build requirement table distinguishing SDK vs toolchain builds.
    - secubox-tools/README.md updated to v1.1.0 with SDK vs toolchain guidance.

11. **Service Registry & HAProxy ACME v0.15.0 (2026-01-28)**
    - `service-registry`: Unified service aggregation dashboard with dynamic health checks, URL readiness wizard, public IP detection, and external port checks.
    - `haproxy`: Webroot ACME mode (no HAProxy restart), async cert workflow, auto-open firewall when publishing.
    - Menu reorganization: CrowdSec, Threat Monitor, Network Diagnostics, WireGuard all moved to LuCI Services menu.
    - `tor-shield`: Exit node hostname (reverse DNS), presets with immediate activation, excluded destinations for CDN/direct, master protection switch.
    - `network-tweaks`: AdGuard Home DNS control, CDN cache and WPAD proxy controls, cumulative impact counters for HAProxy vhosts/LXC/firewall.
    - `client-guardian`: Safe defaults, emergency clear, and safety limits.
    - `metablogizer`: Improved site creation and HAProxy integration.
    - Portal: HTTP health checks and speedtest integration.
    - CrowdSec: Dynamic LAPI port detection.
    - 30 commits, 15 feat / 12 fix / 3 refactor.

12. **App Store KISS Evolution & Dependency Cleanup (2026-01-29)**
    - Renamed `luci-app-secubox-bonus` to `secubox-app-bonus` (feeds-based architecture).
    - Implemented KISS Evolution for app store: feeds, profiles, skills, feedback system.
    - Stripped all libc/libubox/libubus/libuci dependencies from SecuBox packages.
    - Added `PKG_FLAGS:=nonshared` to prevent automatic libc dependency injection.

13. **P2P Hub & SecuBox Console (2026-01-30)**
    - `secubox-p2p`: Full P2P Hub with globe peer visualization, Hub Registry, Services Registry, parallel component sources, auto-self mesh, master deployment, DNS bridge, WireGuard mirror, Gitea repository creation, mesh backup, test cloning, gigogne distribution mode, and mDNS service publishing.
    - `secubox-console`: Linux host TUI frontend with CLI tools lexical reference.
    - `cdn-cache`: Added MITM SSL bump support for HTTPS caching.
    - `metablogizer`: Tor hidden service integration, DNS resolution fixes, permissions fixes.
    - `streamlit`: ZIP upload with selective tree extraction.
    - `crowdsec-dashboard`: Extensible theming system (later removed), UCI ubus permissions.
    - `secubox-core`: P2P Hub API and wizard-first menu.
    - `secubox-app-bonus`: Added `secubox-feed install all` command.
    - 40+ commits — largest single-day effort in project history.

14. **P2P MirrorBox & Factory Dashboard (2026-01-31)**
    - `secubox-p2p` v0.6.0: MirrorBox NetMesh Catalog with DNS federation, distributed mesh services panel, WAN IP and WireGuard tunnel redundancy, mDNS service publishing, REST API for mesh visibility.
    - `secubox-factory`: Unified dashboard with signed Merkle snapshots and HMAC-style signing for OpenWrt compatibility.
    - `portal`: KISS redesign with service categorization.
    - `crowdsec-dashboard`: KISS rewrite, console enrollment, CrowdSec theme integration, dynamic port/path detection.
    - `secubox-swiss`: Unified CLI tool for SecuBox operations.
    - `jitsi`: Jitsi Meet video conferencing integration.
    - `mitmproxy`: HAProxy backend inspection, token auth, enhanced threat detection analytics v2.0.
    - `secubox-core`: P2P mesh API endpoints for console discovery.

15. **KISS UI Rewrites & DNS Guard (2026-02-01)**
    - `streamlit`: KISS UI redesign with instances management, Gitea integration, and multiple upload bug fixes.
    - `metablogizer`: KISS UI redesign with backend status display.
    - `ollama`: KISS UI rewrite with model suggestions and thermal monitoring.
    - `netdiag`: Thermal monitoring integration.
    - `dnsguard`: New DNS Guard app with provider lookup methods.
    - `haproxy`: AdGuard Home detection, improved service discovery, reserved ports with listening verification.
    - `p2p`: Distributed catalog with Gitea sync and health probing.
    - `mitmproxy`: Enhanced threat patterns; moved to Security menu.
    - `network-tweaks`: Moved to Network menu.
    - `crowdsec-dashboard`: Nav path fixes, alerts/countries display fixes.
    - `wireguard-dashboard`: QR code generation fix.
    - `exposure`: Reserved ports with listening verification.

16. **WAF Auto-Ban & Security Hardening (2026-02-02)**
    - `waf`: Sensitivity-based auto-ban system with CrowdSec integration and comprehensive CVE detection patterns (including CVE-2025-15467).
    - `mitmproxy`: WAN protection mode for incoming traffic inspection; LAN transparent proxy disabled by default.
    - `simplex`: SimpleX Chat self-hosted messaging servers.
    - `crowdsec`: KISS setup simplification, CAPI enrollment status, restored working setup page.
    - `local-build`: Added missing toolchain package shorthands and feeds path fix.
    - WAF auto-ban statistics added to dashboards.

17. **Mesh Security & MAC Guardian (2026-02-03)**
    - `mac-guardian`: New WiFi MAC security monitor with DHCP lease protection for odhcpd.
    - `master-link`: Secure mesh onboarding with dynamic join IPK generation.
    - `security-threats`: KISS rewrite with mesh threat intelligence, LXC mitmproxy detection.
    - `p2p`: Decentralized threat intelligence sharing via mesh.
    - `tor-shield`: Server mode for split-routing with public IP preservation.
    - `wireguard-dashboard`: jshn bypass for QR code (argument size limit), peer private key persistence in UCI, server endpoint persistence.
    - `localai`: gte-small preset, RPC expect unwrapping and chat JSON escaping fixes.
    - `lyrion`: WAN access checkbox for firewall rules, networking fixes for device discovery.
    - `tools`: SecuBox image builder and sysupgrade scripts.
    - RPCD/LuCI frontend guidelines added to CLAUDE.md.
    - KISS READMEs added for all 46 remaining packages.

18. **New Packages & Exposure Redesign (2026-02-04)**
    - `jellyfin`: New media server package with LXC container, uninstall/update/backup, HAProxy integration, and LuCI actions.
    - `zigbee2mqtt`: Complete rewrite from Docker to LXC Alpine container.
    - `device-intel`: New device intelligence package with OUI emoji display (BusyBox compatibility fixes, SDK build pattern alignment).
    - `dns-provider`: New DNS provider management package.
    - `exposure`: KISS redesign with enriched service names, vhost integration, DNS domain sorting; toggle switch fix.
    - `streamlit`: Chunked upload to bypass uhttpd 64KB JSON limit, UTF-8 `.py` file upload fix, auto-install requirements from ZIP, non-standard filename support.
    - `crowdsec-dashboard`: Decisions list fix (wrong RPC expect key).
    - RPCD: BusyBox ash `local` keyword compatibility fix (wrap call handlers in function).
    - `glances`: Full host system visibility — LXC bind mounts for `/rom`, `/overlay`, `/boot`, `/srv`, Docker socket at `/run/docker.sock` (symlink loop fix), `@exit_after` fs plugin patch (multiprocessing fails in LXC), host hostname via `lxc.uts.name`, OpenWrt OS identity from `/etc/openwrt_release`, pre-generated `/etc/mtab` from host `/proc/mounts`.
    - `zigbee2mqtt`: Direct `/dev/ttyUSB0` passthrough (socat TCP bridge fails ASH protocol), adapter `ezsp`→`ember` (z2m 2.x), `ZIGBEE2MQTT_DATA` env var, `mosquitto-nossl` dependency.
    - `smbfs`: New SMB/CIFS remote mount manager package — UCI config, `smbfsctl` CLI (add/remove/mount/umount/test/status), auto-mount init script, credentials storage, Jellyfin+Lyrion integration, catalog entry.
    - `jellyfin`: KISS READMEs for both backend and LuCI packages.
    - `domoticz`: Rewrite from Docker to LXC Debian container with native binary from GitHub releases. LuCI dashboard with IoT integration status (Mosquitto, Zigbee2MQTT, MQTT bridge), service lifecycle, HAProxy, mesh P2P, logs. `domoticzctl` with `configure-mqtt` (auto Mosquitto+Z2M bridge), `configure-haproxy`, `backup/restore`, `mesh-register`, `uninstall`. UCI extended with mqtt/network/mesh sections. Catalog updated.
    - LXC cgroup2 fix: Added `lxc.tty.max`, `lxc.pty.max`, `lxc.cgroup2.devices.allow` for standard character devices, and `lxc.seccomp.profile` disable to fix terminal allocation failures on cgroup v2 systems. Applied to `streamlit` and `domoticz`.
    - `metablogizer`: Chunked upload to bypass uhttpd 64KB JSON limit (same pattern as Streamlit). Added `upload_chunk` and `upload_finalize` RPCD methods, binary file support via ArrayBuffer read.
    - `p2p`: P2P App Store Emancipation — decentralized package distribution across mesh peers.
      CGI API: `/api/factory/packages` (local catalog JSON), `/api/factory/packages-sync` (aggregated mesh catalog).
      RPCD: 7 new methods for peer packages, fetch, sync, feed settings.
      CLI: `secubox-feed peers/search/fetch-peer/fetch-any/sync-peers` commands.
      LuCI: `packages.js` view under MirrorBox > App Store with LOCAL/PEER badges, unified catalog, one-click fetch/install.
      UCI: `config p2p_feed` section with share_feed, auto_sync, sync_interval, prefer_local, cache_ttl.
    - `rustdesk`: New self-hosted RustDesk relay server package — pre-built ARM64 binaries from GitHub releases (hbbs/hbbr), auto-key generation, `rustdeskctl` CLI with install/status/keygen/logs/configure-firewall/mesh-register.
    - `guacamole`: New Apache Guacamole clientless remote desktop gateway — LXC Debian container with guacd + Tomcat, UCI-based connection management (SSH/VNC/RDP), `guacamolectl` CLI with install/add-ssh/add-vnc/add-rdp/list-connections/configure-haproxy.
    - `services.js`: Fixed RPC expect unwrapping bug causing empty local services list.
    - `content-pkg`: New content distribution system — `secubox-content-pkg` CLI packages Metablogizer sites and Streamlit apps as IPKs for P2P mesh distribution. Auto-publish hooks in metablogizerctl/streamlitctl. `secubox-feed sync-content` auto-installs content packages from peers. Sites get HAProxy vhosts, Streamlit apps run as service instances.
    - `devstatus.js`: New Development Status widget under MirrorBox > Dev Status — generative/dynamic dashboard with real-time polling, Gitea commit activity (15 recent commits), repository stats, MirrorBox App Store package counts (local/peer/unique), v1.0 progress bar (0-100%) with 8 milestone categories, color-coded completion indicators.

19. **ksmbd & UI Consistency (2026-02-05)**
    - `ksmbd`: New `secubox-app-ksmbd` mesh media server package — `ksmbdctl` CLI with enable/disable/status/add-share/remove-share/list-shares/add-user/mesh-register, UCI config with pre-configured shares (Media, Jellyfin, Lyrion, Backup), Avahi mDNS announcement, P2P mesh registration.
    - `client-guardian`: Ported to `sh-page-header` chip layout with 6 status chips (Online, Approved, Quarantine, Banned, Threats, Zones).
    - `auth-guardian`: Ported to `sh-page-header` chip layout with 4 status chips (Status, Sessions, Portal, Method), sessions table, quick actions card.

20. **Navigation Component Refactoring (2026-02-05)**
    - `secubox/nav.js`: Unified navigation widget with auto-theme initialization.
      - `renderTabs(active)`: Main SecuBox tabs with automatic Theme.init() and CSS loading.
      - `renderCompactTabs(active, tabs, options)`: Compact variant for nested modules.
      - `renderBreadcrumb(moduleName, icon)`: Back-navigation to SecuBox dashboard.
    - Eliminated ~1000 lines of duplicate CSS from module nav files.
    - Updated modules: `cdn-cache`, `client-guardian`, `crowdsec-dashboard`, `media-flow`, `mqtt-bridge`, `system-hub`.
    - Views no longer need to require Theme separately or manually load CSS.

21. **Monitoring UX Improvements (2026-02-05)**
    - Empty-state loading animation for charts during 5-second data collection warmup.
      - Animated "Collecting data..." overlay with pulsing dots.
      - Chart legend shows "Waiting" → "Live" transition.
      - Cyberpunk theme support for empty state styling.
    - Dynamic bandwidth units via new `formatBits()` helper.
      - Network rates now display in bits (Kbps/Mbps/Gbps) instead of bytes.
      - Uses SI units (1000 base) for industry-standard notation.
      - Dash placeholder ("— ↓ · — ↑") before first data point.

22. **Punk Exposure Emancipate CLI (2026-02-05)**
    - `secubox-exposure emancipate <service> <port> <domain> [--tor] [--dns] [--mesh] [--all]`
      - Unified multi-channel exposure: Tor + DNS/SSL + Mesh in single command.
      - Creates DNS A record via `dnsctl`, HAProxy vhost, requests certificate.
      - Publishes to mesh via `secubox-p2p publish`.
      - Stores emancipation state in UCI for status tracking.
    - `secubox-exposure revoke <service> [--tor] [--dns] [--mesh] [--all]`
      - Inverse of emancipate: removes exposure from selected channels.
      - Cleans up DNS records, HAProxy vhosts, certificates, mesh publishing.
    - Enhanced `status` command shows emancipated services with active channels.

23. **Punk Exposure LuCI Dashboard (2026-02-05)**
    - RPCD handler extended with three new methods:
      - `emancipate` - orchestrates multi-channel exposure via CLI
      - `revoke` - removes exposure from selected channels
      - `get_emancipated` - returns list of emancipated services with channel status
    - API wrapper (`exposure/api.js`) exports `emancipate()`, `revoke()`, `getEmancipated()`.
    - ACL updated in `luci-app-exposure.json` for new methods.
    - Dashboard UI enhancements:
      - New Mesh column with toggle switch (blue theme)
      - Emancipate button in header with rocket emoji
      - Multi-channel modal with Tor/DNS/Mesh checkboxes
      - Mesh badge count in header stats
    - CSS additions: `.exp-badge-mesh`, `.mesh-slider`, `.exp-btn-action`.

24. **Jellyfin Post-Install Setup Wizard (2026-02-05)**
    - 4-step modal wizard for first-time Jellyfin configuration.
    - RPCD methods added to `luci.jellyfin`:
      - `get_wizard_status` - checks container state and wizard completion
      - `set_wizard_complete` - marks wizard as finished in UCI
      - `add_media_path` / `remove_media_path` - manage media library entries
      - `get_media_paths` - returns configured media libraries
    - Wizard auto-triggers when installed but `wizard_complete=0`.
    - Steps: Welcome (Docker/container checks), Media (add paths), Network (domain/HAProxy), Complete.
    - New CSS file `jellyfin/wizard.css` with step indicators and form styling.
    - Makefile updated to install CSS resources.

25. **MAC Guardian Feed Integration (2026-02-05)**
    - Built and added `secubox-app-mac-guardian` and `luci-app-mac-guardian` IPKs to bonus feed.
    - Synced `luci-app-mac-guardian` to local-feed (backend was already synced).
    - Updated `apps-local.json` catalog with proper metadata:
      - `luci-app-mac-guardian`: category "security", icon "wifi", description "WiFi MAC address security monitor with spoofing detection"
      - `secubox-app-mac-guardian`: icon "wifi", description "WiFi MAC security backend with CrowdSec integration"
    - Package features: MAC spoofing detection, OUI anomaly detection, MAC floods, CrowdSec scenarios integration.

26. **Fanzine v3 Roadmap Alignment (2026-02-06)**
    - Restructured TODO.md and WIP.md to align with SecuBox Fanzine v3 4-layer architecture:
      - **Couche 1 — Core Mesh**: 35+ modules, v0.18 priorities, testing/validation, CVE Layer 7
      - **Couche 2 — AI Gateway**: Data Classifier, 6 Autonomous Agents, MCP Server, provider hierarchy
      - **Couche 3 — MirrorNetworking**: EnigmaBox → MirrorNet, dual transport, Services Mirrors, VoIP/Matrix
      - **Couche 4 — Roadmap**: v0.18/v0.19/v1.0/v1.1+ milestones, certifications (ANSSI, ISO, NIS2)
    - Added strategic reference to Fanzine v3 document.
    - Consolidated completed items under "Resolved" section.
    - Created version milestone checklists for tracking progress.

27. **LocalAI Upgrade to v3.9.0 (2026-02-06)**
    - Upgraded `secubox-app-localai` from v2.25.0 to v3.9.0.
    - New features in v3.9.0:
      - **Agent Jobs Panel**: Schedule and manage background agentic tasks via web UI and API
      - **Memory Reclaimer**: LRU eviction for loaded models, automatic VRAM cleanup
      - **VibeVoice backend**: New voice synthesis support
    - Updated README with complete CLI reference, model presets table, API endpoints.
    - Part of v0.18 AI Gateway roadmap (Couche 2).

28. **MCP Server Implementation (2026-02-06)**
    - Created `secubox-mcp-server` package — Model Context Protocol server for AI integration.
    - **Protocol**: JSON-RPC 2.0 over stdio, MCP version 2024-11-05.
    - **Core tools** (9 total):
      - `crowdsec.alerts`, `crowdsec.decisions` — CrowdSec threat intelligence
      - `waf.logs` — WAF/mitmproxy threat events
      - `dns.queries` — DNS statistics from AdGuard Home/dnsmasq
      - `network.flows` — Network traffic summary with interface stats
      - `system.metrics` — CPU, memory, disk, temperature monitoring
      - `wireguard.status` — VPN tunnel status with peer details
      - `uci.get`, `uci.set` — OpenWrt configuration access (set disabled by default)
    - **AI-powered tools** (5 total, require LocalAI):
      - `ai.analyze_threats` — AI analysis of CrowdSec alerts with recommendations
      - `ai.cve_lookup` — CVE vulnerability analysis with mitigation advice
      - `ai.suggest_waf_rules` — AI-suggested mitmproxy/WAF filter patterns
      - `ai.explain_ban` — Explain CrowdSec ban decisions in plain language
      - `ai.security_posture` — Full security assessment with scoring
    - **Security features**:
      - UCI-based tool whitelist — only allowed tools can be invoked
      - Sensitive data blocked in uci.get (password, secret, key, token)
      - uci.set disabled by default, requires explicit enable
      - Data classification support (local_only, sanitized, cloud_direct)
    - **Claude Desktop integration** via SSH:
      ```json
      {"mcpServers":{"secubox":{"command":"ssh","args":["root@192.168.255.1","/usr/bin/secubox-mcp"]}}}
      ```
    - Files: `secubox-mcp` main server, `protocol.sh` JSON-RPC handler, 8 tool modules.
    - Part of v0.18 AI Gateway roadmap (Couche 2).

29. **Threat Analyst Agent Implementation (2026-02-05)**
    - Created `secubox-threat-analyst` — AI-powered autonomous threat analysis and filter generation agent.
    - **Architecture**:
      - Collector: Gathers threats from CrowdSec, mitmproxy, netifyd DPI
      - Analyzer: LocalAI-powered intelligent analysis and pattern recognition
      - Generators: Rule creation for three targets
      - Appliers: Auto-apply or queue for approval
    - **Generated rule types**:
      - `mitmproxy`: Python filter class with IP blocklist, URL patterns, User-Agent detection
      - `CrowdSec`: YAML scenarios for AI-detected attack patterns
      - `WAF`: JSON rules for SQLi, XSS, path traversal, scanner detection
    - **CLI commands**: status, run, daemon, analyze, generate, gen-mitmproxy, gen-crowdsec, gen-waf, list-pending, approve, reject
    - **UCI configuration**: interval, LocalAI URL/model, auto-apply per target (mitmproxy auto, CrowdSec/WAF queued), min_confidence, max_rules_per_cycle
    - Created `luci-app-threat-analyst` — LuCI dashboard with AI chatbot.
    - **Dashboard features**:
      - Status panel: daemon state, LocalAI connectivity, threat counts
      - AI Chat: real-time conversation with threat analyst AI
      - Pending rules: approve/reject queue for generated rules
      - Threats table: recent security events with severity badges
    - **RPCD methods**: status, get_threats, get_alerts, get_pending, chat, analyze, generate_rules, approve_rule, reject_rule, run_cycle
    - Part of v0.18 AI Gateway roadmap (Couche 2).

30. **DNS Guard AI Migration (2026-02-06)**
    - Created `secubox-dns-guard` — AI-powered DNS anomaly detection daemon.
    - **Detection modules** (5 total):
      - `dga`: Domain Generation Algorithm detection via Shannon entropy analysis (threshold 3.2)
      - `tunneling`: DNS tunneling/exfiltration detection (subdomain length, base64/hex patterns, TXT rate)
      - `rate_anomaly`: Unusual query rate detection (queries/min, unique domains/min thresholds)
      - `known_bad`: Known malicious domain matching against external blocklists
      - `tld_anomaly`: Suspicious TLD detection (xyz, top, club, etc.) and punycode/IDN homograph detection
    - **LocalAI integration**:
      - Intelligent threat analysis and domain classification (BLOCK/MONITOR/SAFE)
      - Pattern analysis and malware family identification
      - Single domain analysis via CLI
    - **Approval workflow**:
      - Auto-apply mode for trusted detections
      - Queue mode for human approval (configurable per confidence threshold)
      - Pending blocks approval via CLI or LuCI
    - **CLI commands**: status, run, daemon, analyze, detect, check <domain>, stats, top-domains, top-clients, list-pending, approve/reject/approve-all
    - **UCI configuration**: interval, LocalAI URL/model, auto_apply_blocks, min_confidence (80%), max_blocks_per_cycle, per-detector settings
    - Updated `luci-app-dnsguard` to v1.1.0:
      - New "AI Guard" tab with daemon toggle, alert/pending/blocked counts
      - Pending blocks approval panel with approve/reject actions
      - Real-time alerts panel with type-colored badges
      - "Analyze" tab with domain checker and detection module status
      - RPCD extended with 11 new methods: guard_status, get_alerts, get_pending, approve_block, reject_block, approve_all, ai_check, get_blocklist, unblock, get_stats, toggle_guard
    - Part of v0.18 AI Gateway roadmap (Couche 2).

31. **LocalAI Multi-Channel Emancipation (2026-02-06)**
    - Exposed LocalAI (port 8091) via Punk Exposure system with 3 channels:
      - **Tor**: `b7lmlfs3b55jhgqdwbn6unhjhlfflq6ch235xa2gsdvxe7toxcf7qyad.onion`
      - **DNS/SSL**: `localai.secubox.local` via HAProxy with ACME certificate
      - **mDNS**: `_secubox._tcp.local` mesh advertisement via Avahi
    - Command: `secubox-exposure emancipate localai 8091 localai.secubox.local --all`
    - Documented MirrorNetworking vision for v0.19:
      - Master/slave hierarchical domain delegation (*.sb → xxx.sb)
      - Service mirroring via reverse proxy chaining
      - Gossip-based exposure config sync
      - Submastering/multimixslaving architecture

32. **Threat Analyst KISS Dashboard v0.1.0 (2026-02-05)**
    - Regenerated `luci-app-threat-analyst` following CrowdSec dashboard KISS template pattern.
    - **Architectural changes**:
      - `api.js`: Migrated from plain object to `baseclass.extend()` pattern
      - `dashboard.css`: External CSS file (loaded dynamically in view)
      - `dashboard.js`: View-only JS following CrowdSec pattern with `view.extend()`
    - **CVE integration**:
      - System Health: New "CVE Alerts" indicator with warning icon (yellow) when CVEs detected
      - Threats table: New CVE column with hyperlinks to NVD (`https://nvd.nist.gov/vuln/detail/CVE-XXXX-XXXXX`)
      - CVE extraction: `extractCVE()` function in API parses CVE-YYYY-NNNNN patterns from scenarios
      - CVE row styling: Red-tinted background for CVE-related threats
    - **RPCD updates**:
      - Status method now returns `cve_alerts` count from CrowdSec alerts
      - Fixed output bug (grep `|| echo 0` causing double output)
    - **CSS additions**:
      - `.ta-health-icon.warning` for CVE alerts in health section
      - `.ta-cve-link` for NVD hyperlinks (red badge style)
      - `.ta-cve-row` for highlighted CVE threat rows
    - Following LuCI UI Generation Model Template v0.1.0 for future KISS modules.

33. **Unified Backup Manager & Custom Mail Server (2026-02-05)**
    - Created `secubox-app-backup` — unified backup system for LXC containers, UCI config, service data.
      - **CLI commands**: create (full/config/containers/services), list, restore, status, cleanup
      - **Container ops**: container list/backup/restore/backups
      - **Profile ops**: profile list/create/apply/share (delegates to secubox-profile)
      - **Remote sync**: sync --push/--pull (Gitea integration)
      - **Libraries**: containers.sh, config.sh, remote.sh
      - **Storage structure**: /srv/backups/{config,containers,services,profiles}
    - Created `luci-app-backup` — LuCI dashboard for backup management.
      - **Status panel**: storage path, usage, last backup times
      - **Quick actions**: Full/Config/Containers backup buttons
      - **Container table**: name, state, size, backup count, backup button
      - **Backup history**: file, type, size, date (sorted by timestamp)
      - **RPCD methods**: status, list, container_list, create, restore, cleanup, container_backup, container_restore
    - Created `secubox-app-mailserver` — custom Postfix + Dovecot mail server in LXC container.
      - **mailctl CLI**: install, start/stop/restart, status
      - **User management**: user add/del/list/passwd, alias add/list
      - **SSL**: ssl-setup (ACME DNS-01), ssl-status
      - **DNS integration**: dns-setup (creates MX, SPF, DMARC via dnsctl)
      - **Mesh backup**: mesh backup/restore/sync/add-peer/peers/enable/disable
      - **Webmail integration**: webmail status/configure (Roundcube container)
      - **Libraries**: container.sh, users.sh, mesh.sh
    - Enhanced `dnsctl` with subdomain generation and mail DNS:
      - `generate <service> [prefix]` — auto-create subdomain A record with public IP
      - `suggest [category]` — subdomain name suggestions (web, mail, dev, media, iot, security)
      - `mail-setup [host] [priority]` — create MX, SPF, DMARC records
      - `dkim-add [selector] <pubkey>` — add DKIM TXT record
    - Renamed `secbx-webmail` Docker container to `secubox-webmail` for consistency.

34. **HAProxy/Mailserver LXC cgroup Fixes & Documentation (2026-02-06)**
    - Fixed HAProxy LXC container cgroup mount failure:
      - Removed `lxc.mount.auto = proc:mixed sys:ro cgroup:mixed` which fails on cgroup v2 hosts
      - Simplified to explicit `lxc.mount.entry` bind mounts only
      - Updated `haproxyctl` `lxc_create_config()` function with working config
    - Fixed Docker-to-LXC mailserver connectivity:
      - Added socat TCP proxies on ports 10143/10025 in mailserver init.d script
      - Configured Dovecot with `disable_plaintext_auth = no` for local connections
      - Roundcube can now reach LXC mailserver via host-bridged ports
    - Documentation updates:
      - Added "LXC container fails with cgroup:mixed" section to FAQ-TROUBLESHOOTING.md
      - Updated CLAUDE.md Session Startup section to include FAQ-TROUBLESHOOTING.md consultation
      - Key recommendation: avoid `lxc.mount.auto` entirely, use explicit bind mounts

35. **Vortex DNS - Meshed Subdomain Delegation (2026-02-05)**
    - Created `secubox-vortex-dns` — meshed multi-dynamic subdomain delegation system.
    - **Modes**:
      - **Master**: Owns wildcard domain (*.secubox.io), delegates subzones to slaves
      - **Slave**: Receives delegated subdomain from master (node1.secubox.io)
      - **Submaster**: Hierarchical delegation (master → submaster → slaves)
      - **Standalone**: Default mode, mesh-only participation
    - **CLI commands** (`vortexctl`):
      - Master: `master init <domain>`, `master delegate <node> <zone>`, `master revoke <zone>`, `master list-slaves`
      - Slave: `slave join <master> <token>`, `slave leave`, `slave status`
      - Mesh: `mesh sync`, `mesh publish <service> <domain>`, `mesh unpublish`, `mesh status`
      - Submaster: `submaster promote`, `submaster demote`
      - General: `status`, `daemon`
    - **Mesh integration**:
      - First Peek: Auto-registers new services in mesh DNS
      - Gossip-based exposure config sync via `secubox-p2p`
      - Published services tracked in `/var/lib/vortex-dns/published.json`
    - **DNS provider integration**:
      - Uses `dnsctl` from `secubox-app-dns-provider` for programmatic DNS record management
      - Auto-creates wildcard A record on master init
      - NS/A records for zone delegation
    - Created `luci-app-vortex-dns` — LuCI dashboard.
      - Status panel: mode badge, enabled state, sync interval, last sync time
      - Master section: wildcard domain, DNS provider, delegated slave count, zones table
      - Slave section: parent master, delegated zone
      - Mesh section: gossip state, First Peek, peer count, published services
      - Actions: Sync Mesh, Initialize as Master, Join as Slave, Delegate Zone
    - **RPCD methods**: status, get_slaves, get_peers, get_published, master_init, delegate, revoke, slave_join, mesh_sync, mesh_publish
    - Part of v0.19 MirrorNetworking roadmap (Couche 3).

36. **Network Anomaly Detection Agent (2026-02-06)**
    - Created `secubox-network-anomaly` — AI-powered network traffic anomaly detection.
    - **Detection modules** (5 total):
      - `bandwidth_anomaly`: Traffic spike detection via EMA baseline comparison
      - `connection_flood`: Connection count threshold monitoring
      - `port_scan`: Unique destination port enumeration detection
      - `dns_anomaly`: DNS query volume anomaly detection
      - `protocol_anomaly`: TCP/UDP ratio deviation (flags >50% UDP as suspicious)
    - **Data collection**:
      - Interface bandwidth from `/sys/class/net/*/statistics/`
      - Connection tracking from `/proc/net/nf_conntrack`
      - DNS queries from dnsmasq/AdGuard logs
    - **CLI commands** (`network-anomalyctl`):
      - `status`, `run`, `daemon` — service control
      - `analyze` — LocalAI-powered threat assessment
      - `list-alerts`, `ack <id>`, `clear-alerts` — alert management
      - `baseline [reset]` — EMA baseline control
    - **UCI configuration**:
      - Thresholds: bandwidth_spike_percent (200%), new_connections_per_min (50), unique_ports_per_host (20), dns_queries_per_min (100)
      - Detection flags: per-detector enable/disable
      - LocalAI integration: url, model, min_confidence (75%)
      - Auto-block: optional CrowdSec integration
    - Created `luci-app-network-anomaly` — LuCI dashboard.
      - Status panel: daemon state, LocalAI, alert count, connection count
      - Health checks: daemon, LocalAI, auto-block, interval, last run
      - Network stats: real-time RX/TX, connections, unique ports
      - Actions: Run Detection, AI Analysis, Reset Baseline, Clear Alerts
      - Alerts table: time, type, severity, message, ack button
    - **RPCD methods**: status, get_alerts, get_stats, run, ack_alert, clear_alerts, reset_baseline, analyze
    - Part of v0.19 AI Gateway roadmap (Couche 2).

37. **LocalRecall AI Memory System (2026-02-06)**
    - Created `secubox-localrecall` — persistent memory for AI agents.
    - **Memory categories**:
      - `threats`: Security threat patterns and detections
      - `decisions`: Agent decisions with outcomes (approved/rejected/auto)
      - `patterns`: Learned behavioral patterns
      - `configs`: Configuration snapshots and changes
      - `conversations`: AI conversation context
    - **Memory storage**:
      - JSON-based storage in `/var/lib/localrecall/memories.json`
      - EMA-based importance scoring (1-10)
      - Access tracking with timestamps and counts
      - Category-based indexing
    - **CLI commands** (`localrecallctl`):
      - `status`, `add`, `get`, `search`, `list`, `recent`, `important`
      - `delete`, `cleanup`, `export`, `import`
      - `summarize`, `context`, `stats`
    - **LocalAI integration**:
      - `summarize_memories()` — AI-powered memory summarization
      - `auto_memorize()` — Extract key facts from text
      - `get_agent_context()` — Build context for agent tasks
      - `record_decision()`, `record_threat()` — Structured memory helpers
    - **UCI configuration**:
      - Retention: max_memories (1000), retention_days (90)
      - Categories: enable/disable per category
      - Agents: enable/disable per agent
      - Cleanup: auto_cleanup, cleanup_hour, keep_important
    - Created `luci-app-localrecall` — LuCI dashboard.
      - Stats: total/threats/decisions/patterns counts
      - Categories panel with icons and counts
      - Agent breakdown panel
      - Actions: AI Summary, Search, Cleanup, Export
      - Add memory form with category, importance, content
      - Recent memories table with delete
    - **RPCD methods**: status, get_memories, search, stats, add, delete, cleanup, summarize, export, import
    - Part of v0.19 AI Gateway roadmap (Couche 2).

38. **AI Insights Dashboard (2026-02-06)**
    - Created `luci-app-ai-insights` — unified AI security insights dashboard.
    - **Security Posture Score**:
      - 0-100 score with color-coded display (Excellent/Good/Fair/Poor/Critical)
      - Dynamic factor calculation: LocalAI status, agent online counts, CrowdSec alerts, CVE severity
      - Real-time score updates via polling
    - **Agent Status Grid**:
      - Visual cards for 4 agents: Threat Analyst, DNS Guard, Network Anomaly, CVE Triage
      - Online/offline status with color indicators
      - Alert count badges per agent
    - **Aggregated Alerts**:
      - Unified view of alerts from all agents
      - Source-colored badges (rule/alert/cve)
      - Relative timestamps
    - **Actions**:
      - Run All Agents — triggers detection cycles on all agents
      - AI Analysis — LocalAI-powered security assessment with recommendations
      - View Timeline — security events from system log (24h)
      - Link to LocalRecall memory dashboard
    - **RPCD methods**: status, get_alerts, get_posture, get_timeline, run_all, analyze
    - Part of v0.19 AI Gateway roadmap (Couche 2).

39. **MirrorNet Core Packages (2026-02-07)**
    - Created `secubox-mirrornet` — mesh orchestration core with 5 library modules.
    - **Identity module** (`identity.sh`):
      - DID generation: `did:plc:<16-char-fingerprint>` (AT Protocol compatible)
      - HMAC-SHA256 keypair management with Ed25519 fallback
      - Key rotation with backup, identity document export/import
      - Peer identity storage and resolution
    - **Reputation module** (`reputation.sh`):
      - Trust scoring (0-100) with decay and ban thresholds
      - Event logging: sync_success/failed, valid/invalid_ioc, fast/slow_response, offline/online
      - Trust levels: excellent (80+), good (60+), moderate (40+), low (20+), untrusted
      - Ban threshold (default 10), min_trust threshold (default 20)
    - **Mirror module** (`mirror.sh`):
      - Service mirroring via reverse proxy chaining
      - Upstream management with priority-based failover
      - HAProxy backend configuration generation
      - Health check integration with automatic failover
    - **Gossip module** (`gossip.sh`):
      - Enhanced gossip protocol with priority routing (critical > high > normal > low > background)
      - TTL-based message forwarding with configurable max_hops (default 5)
      - Deduplication with 5-minute window
      - Message types: ioc, peer_status, config_sync, service_announce, mirror_update, reputation_update
    - **Health module** (`health.sh`):
      - Per-peer latency and packet loss monitoring
      - HTTP health checks with configurable endpoints
      - Anomaly detection against EMA baselines
      - Alert generation with acknowledgment workflow
    - **CLI** (`mirrorctl`): 30+ commands for identity, reputation, mirror, gossip, health, daemon
    - **UCI configuration**: roles (master/submaster/peer), gossip interval, health thresholds, mirror settings
    - Created `luci-app-secubox-mirror` — LuCI dashboard.
      - Identity card: DID, hostname, role, version
      - Status grid: peers, messages, services, alerts
      - Peer reputation table with trust levels and reset action
      - Gossip stats: sent/received/forwarded/dropped
      - Health alerts with acknowledgment
      - Mirrored services table
    - **RPCD methods**: status, get_identity, get_peers, get_reputation, get_health, get_mirrors, get_gossip_stats, get_alerts, reset_reputation, ack_alert, add_mirror, trigger_failover, broadcast
    - Part of v0.19 MirrorNetworking roadmap (Couche 3).

40. **SecuBox Identity Package (2026-02-07)**
    - Created `secubox-identity` — standalone DID identity management.
    - **Core module** (`core.sh`):
      - DID generation: `did:plc:<fingerprint>` from machine-id + MAC
      - Identity document creation (DID Document format with @context)
      - Peer identity import/export
      - Identity backup and restore
    - **Keys module** (`keys.sh`):
      - HMAC-SHA256 keypair generation (Ed25519 fallback if available)
      - Key rotation with configurable backup
      - Sign/verify operations
      - Key rotation check (configurable rotation_days: default 90)
    - **Trust module** (`trust.sh`):
      - Peer trust scoring (0-100)
      - Trust events: valid/invalid_signature, successful/failed_exchange, verified_identity, referred_by_trusted
      - Trust levels: verified, trusted, neutral, suspicious, untrusted
      - Ban functionality
    - **CLI** (`identityctl`): 25+ commands for DID, keys, peers, trust, backup
    - **UCI configuration**: did_method, key algorithm, rotation settings, trust thresholds

41. **P2P Intel Package (2026-02-07)**
    - Created `secubox-p2p-intel` — signed IOC sharing for mesh.
    - **Collector module** (`collector.sh`):
      - Source integrations: CrowdSec, mitmproxy, WAF, DNS Guard
      - Severity classification: critical, high, medium, low
      - Scenario-based severity mapping
    - **Signer module** (`signer.sh`):
      - Cryptographic signing of individual IOCs and batches
      - Batch hash verification (SHA256)
      - Identity integration for signer DID
    - **Validator module** (`validator.sh`):
      - Source trust verification (min_source_trust threshold)
      - Age validation (max_age_hours: default 168)
      - Format validation (IP, domain, URL, hash)
      - Local IP whitelist protection
    - **Applier module** (`applier.sh`):
      - Application methods: nftables (ipset), iptables, CrowdSec
      - Ban duration configuration (default 24h)
      - Approval workflow: auto-apply or queue for manual review
      - Pending queue management (approve/reject)
    - **CLI** (`p2p-intelctl`): 20+ commands for collect, sign, share, validate, apply, approve
    - **UCI configuration**: sources enable/disable, signing, validation settings, application method, auto-apply
    - **Daemon**: Configurable collect_interval (default 300s), auto_collect, auto_share, auto_apply
    - Part of v0.19 MirrorNetworking roadmap (Couche 3).

42. **Config Advisor - ANSSI CSPN Compliance (2026-02-07)**
    - Created `secubox-config-advisor` — security configuration analysis and hardening tool.
    - **ANSSI CSPN compliance framework**:
      - 7 check categories: network, firewall, authentication, encryption, services, logging, updates
      - 25+ security check rules with severity levels (critical, high, medium, low, info)
      - JSON rules database in `/usr/share/config-advisor/anssi-rules.json`
    - **Security check modules** (`checks.sh`):
      - Network: IPv6, management access restriction, SYN flood protection
      - Firewall: default deny policy, drop invalid packets, WAN port exposure
      - Authentication: root password, SSH key auth, SSH password auth
      - Encryption: HTTPS enabled, WireGuard configured, DNS encryption
      - Services: CrowdSec running, services bound to localhost
      - Logging: syslog enabled, log rotation configured
    - **Risk scoring module** (`scoring.sh`):
      - 0-100 score with severity weights (critical=40, high=25, medium=20, low=10, info=5)
      - Grade calculation (A-F) based on thresholds (90/80/70/60)
      - Risk level classification: critical, high, medium, low, minimal
      - Score history tracking and trend analysis
    - **ANSSI compliance module** (`anssi.sh`):
      - Compliance rate calculation (percentage of passing rules)
      - Report generation in text, JSON, and Markdown formats
      - Category filtering and strict mode
    - **Remediation module** (`remediate.sh`):
      - Auto-remediation for 7 checks: NET-002, NET-004, FW-001, FW-002, AUTH-003, CRYPT-001, LOG-002
      - Safe vs manual remediation separation
      - Dry-run mode for preview
      - LocalAI integration for AI-powered suggestions
      - Pending approvals queue
    - **CLI** (`config-advisorctl`):
      - Check commands: `check`, `check-category`, `results`
      - Compliance commands: `compliance`, `compliance-status`, `compliance-report`, `is-compliant`
      - Scoring commands: `score`, `score-history`, `score-trend`, `risk-summary`
      - Remediation commands: `remediate`, `remediate-dry`, `remediate-safe`, `remediate-pending`, `suggest`
      - Daemon mode with configurable check interval
    - Created `luci-app-config-advisor` — LuCI dashboard.
      - Dashboard: score circle, grade, risk level, compliance rate, last check time
      - Check results table with status icons
      - Score history table
      - Compliance view: summary cards, progress bar, results by category
      - Remediation view: quick actions, failed checks with apply buttons, pending approvals
      - Settings: framework selection, scoring weights, category toggles, LocalAI config
    - **RPCD methods**: status, results, score, compliance, check, pending, history, suggest, remediate, remediate_safe, set_config
    - **UCI configuration**: main (enabled, check_interval, auto_remediate), compliance (framework, strict_mode), scoring (passing_score, weights), categories (enable/disable), localai (url, model)
    - Part of v1.0.0 certification roadmap (ANSSI CSPN compliance tooling).

43. **Mail Server Port Fixes & Password Reset (2026-02-07)**
    - Fixed mail ports 587 (Submission), 465 (SMTPS), and 995 (POP3S) not listening.
    - **Root causes identified**:
      - Postfix master.cf missing submission and smtps service entries
      - Dovecot 10-master.conf had pop3s listener commented out
      - `dovecot-pop3d` package not installed in Alpine LXC container
    - **mailctl fix-ports command**:
      - Adds submission (587) service to Postfix master.cf with SASL auth
      - Adds smtps (465) service with TLS wrapper mode
      - Installs `dovecot-pop3d` if missing
      - Uncomments pop3/pop3s listeners in Dovecot 10-master.conf
      - Enables SSL on pop3s (995) and imaps (993) listeners
      - Restarts Postfix and Dovecot to apply changes
    - **LuCI password reset feature**:
      - Added "Reset Password" button in mail users table
      - Modal dialog with password and confirmation fields
      - RPCD `user_passwd` method with stdin JSON fallback
      - `callUserPasswd` RPC declaration in overview.js
    - **LuCI Fix Ports button**:
      - Added to Quick Actions section
      - RPCD `fix_ports` method wrapping CLI command
      - Visual feedback with modal spinner
    - Updated container.sh to include `dovecot-pop3d` in initial package list.

44. **MetaBlogizer KISS ULTIME MODE (2026-02-07)**
    - Added `metablogizerctl emancipate <name>` — one-command full exposure workflow.
    - **Workflow steps** (automated in sequence):
      - DNS Registration: Creates A record via `dnsctl` (Gandi/OVH based on availability)
      - Vortex Mesh: Publishes to mesh via `vortexctl mesh publish`
      - HAProxy: Creates backend, server, and vhost with SSL/ACME enabled
      - SSL Certificate: Requests ACME cert via `haproxyctl cert add` (webroot mode)
      - Zero-downtime Reload: Applies HAProxy config via SIGUSR2
    - **Helper functions**:
      - `_emancipate_dns()`: Public IP detection, subdomain extraction, dnsctl integration
      - `_emancipate_vortex()`: Mesh publication if vortex-dns enabled
      - `_emancipate_haproxy()`: UCI backend/server/vhost creation, haproxyctl generate
      - `_emancipate_ssl()`: ACME certificate request with status feedback
      - `_emancipate_reload()`: Graceful HAProxy reload with restart fallback
    - **Usage**: `metablogizerctl create myblog blog.example.com && metablogizerctl emancipate myblog`
    - **Tracking**: Stores `emancipated=1` and `emancipated_at` timestamp in UCI
    - Part of Punk Exposure architecture (multi-channel emancipation).