secubox-openwrt/.claude/WIP.md
CyberMind-FR 70056e02ed feat(ai-gateway): Add /login command with credential validation
- CLI: aigatewayctl login [provider] - validates credentials before saving
- Rollback on auth failure (preserves previous credentials)
- Format warnings for provider-specific API key patterns
- RPCD: login method for LuCI frontend integration
- ACL: Added write permission for login method

docs: Refactor WIP.md and update HISTORY.md
- WIP.md: 1470 → 108 lines (keep only March 2026 items)
- HISTORY.md: Add entries #74-75 (Feb 2026 milestones)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-03-06 09:49:46 +01:00

109 lines
3.4 KiB
Markdown

# Work In Progress (Claude)
_Last updated: 2026-03-06 (AI Gateway Login)_
> **Architecture Reference**: SecuBox Fanzine v3 — Les 4 Couches
---
## Recently Completed
### 2026-03-06
- **AI Gateway `/login` Command**
- CLI: `aigatewayctl login [provider]` - Interactive or scripted provider authentication
- Validates credentials against provider API before saving
- Rollback on validation failure (preserves previous credentials)
- Format warnings: Claude keys should start with `sk-ant-`, OpenAI with `sk-`
- RPCD: `login` method for LuCI frontend integration
- ACL: Added write permission for `login` method
### 2026-03-04
- **SBOM Pipeline for CRA Annex I Compliance**
- `scripts/check-sbom-prereqs.sh` - Prerequisites validation
- `scripts/sbom-generate.sh` - Multi-source SBOM generation
- `scripts/sbom-audit-feed.sh` - PKG_HASH/PKG_LICENSE feed audit
- `.github/workflows/sbom-release.yml` - GitHub Actions with CVE gating
- `SECURITY.md` - CRA Art. 13 §6 compliant vulnerability disclosure
- **AI Gateway Full-Stack Implementation**
- 3-tier data classification: LOCAL_ONLY, SANITIZED, CLOUD_DIRECT
- Provider routing: LocalAI > Mistral EU > Claude > OpenAI > Gemini > xAI
- `aigatewayctl` CLI with classify/sanitize/provider/audit commands
- `luci-app-ai-gateway` with 4 KISS-themed views
### 2026-03-03
- **Comprehensive Service Audit**
- WAF Enforcement: Disabled `waf_bypass` on 21 vhosts
- Container Autostart: Enabled on 9 essential containers
- Glances Fix: Resolved cgroup mount issue
- 18 LXC Containers Running
- **Vortex DNS Firewall Phases 1-4**
- Threat intel aggregator, SQLite blocklist, dnsmasq integration
- HTTP/HTTPS sinkhole server for infected client detection
- DNS Guard AI detection integration
- Mesh threat sharing via secubox-p2p blockchain
- **Image Builder Validation**
- Validated `secubox-image.sh`, `secubox-sysupgrade.sh`
- Fixed curl redirect issue: Added `-L` flag
### 2026-03-02
- **Reverse MWAN WireGuard v2 - Phase 2**
- LuCI Dashboard for Mesh Uplinks
- 9 RPC methods for uplink management
- 10-second live polling
### 2026-03-01
- **Reverse MWAN WireGuard v2 - Phase 1**
- WireGuard mesh peers as backup internet uplinks via mwan3 failover
- `wgctl` CLI: uplink list/add/remove/status/test/failover
- UCI config for global and per-uplink settings
- **Nextcloud Integration Enhancements**
- WAF-safe SSL routing, scheduled backups, SMTP integration
- CalDAV/CardDAV/WebDAV connection info display
---
## In Progress
- **Vortex DNS** - Meshed multi-dynamic subdomain delegation (DONE 2026-02-05)
- `secubox-vortex-dns` package with `vortexctl` CLI
- Master/slave hierarchical DNS delegation
- Wildcard domain management
---
## Next Up
### v1.1+ Extended Mesh
1. **WAF Auto-Ban Tuning** (optional, as-needed)
- Sensitivity threshold adjustment based on production traffic
### Backlog
- SSMTP / mail host / MX record management (v2)
---
## Strategic Documents
- `SecuBox_LocalAI_Strategic_Analysis.html` — AI Management Layer roadmap
- `SecuBox_AI_Gateway_Hybrid_Architecture.html` — Hybrid Local/Cloud architecture
- `SecuBox_MirrorNetworking_Paradigm_Reversal.html` — EnigmaBox autopsy → MirrorNet
- `SecuBox_Fanzine_v3_Feb2026.html` — 4-layer architecture overview
---
## Blockers / Risks
- No automated regression tests for LuCI views; manual verification required after SCP deploy.
- Guacamole ARM64 pre-built binaries not readily available.