Hitonabi
c48e583790
Feat: Vulkan/RADV-Engine + vocab-gepruefte Spec-Drafts + Provisioning/Sync
...
Engine-Cutover ROCm/HIP -> Vulkan/RADV (gfx1151): +12-22% tg auf MoE (llama-bench
verifiziert, fast 53->65 t/s). ROCm-Build bleibt als Rollback unter /opt/llamacpp.
- Backend: vocab-aware Speculative Decoding. services/gguf_meta.py liest den
Tokenizer-Fingerprint (model/pre/n_vocab) direkt aus dem GGUF-Header (ohne Modell-Load);
register_model + migrate_config haengen nur VOCAB-KOMPATIBLE Drafts an (inkl. --spec-type,
das in dieser llama.cpp-Generation noetig ist). Neue Endpoints /api/models/drafts + /{id}/draft.
- Frontend: idiotensichere Spec-Draft-UI (SpecDraftModal) - nur kompatible Drafts waehlbar,
inkompatible gesperrt mit Begruendung; SPEC/SPEC?-Badge nach echtem Aktiv-Status; Rolle in AddModel.
- maintenance.py: Engine-Update-Quelle -> ggml-org/llama.cpp (Build-Nummer-Vergleich),
ENGINE_PATH=/opt/llamacpp-vulkan.
- Startup-Warmup der brains (deploy/warmup.sh, self-detaching ExecStartPost) + deploy/provision-engine.sh.
- Cleanup: tote LiteLLM gateway/config.yaml + alle Referenzen (config.py/backup.py/backup.sh) entfernt;
README + docs/memory aktualisiert.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-27 01:54:29 +02:00
Hitonabi
f1cbfa8e67
docs(2.0): Box-Deploy live (:9001) + Stand/Naechste-Schritte aktualisiert
...
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-25 08:35:39 +02:00
Hitonabi
1f4c987652
feat(2.0): Phase 5 — Betrieb, Observability + Politur
...
Backup: services/backup.py (Memory-SQLite inkl. WAL/SHM + Configs, retain 7)
+ deploy/backup.sh + POST /api/system/backup + GET /api/system/backups.
Services-Health: GET /api/system/services (aggregierte Reachability +
Observability-Links). Frontend: SystemView mit Dienste-Health + Backup-
Button + Links; Theme-Toggle (Hell/Dunkel, persistent).
Lokal verifiziert: Backup-Snapshot + services-Aggregat (TestClient),
Frontend-Build + Browser (Light-Mode bestaetigt via computed styles).
Damit Phasen 0-5 lokal fertig. Docs aktualisiert.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-25 08:29:27 +02:00
Hitonabi
c773dd7eae
feat(2.0): Phase 4 — Hermes-Schicht (MC-Seite + Runbook)
...
MC-Seite: services/agent.py + routers/agent.py (GET /api/agent/status:
Gateway/WebUI-Reachability + Verdrahtungs-Hinweise), AgentView (Status-
Tiles, Hermes-oeffnen-Button, Offline-Hinweis). deploy/hermes-webui.service
(systemd-USER, :8787). Box-Runbook docs/HERMES_SETUP.md (hermes-webui
installieren, Brain=model:auto via Gateway, Tools/MCP verdrahten inkl.
mcp_mc+mcp_memory, SSH-Windows, LiteLLM-Caveat #26489 ).
Lokal verifiziert: agent/status + Frontend-Build + Browser (Hermes-View).
Box-Ausfuehrung der Verdrahtung steht aus (Runbook). Docs aktualisiert.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-25 08:22:28 +02:00
Hitonabi
81468df9c0
feat(2.0): Phase 3 — Memory + MCP
...
Geteiltes Gedaechtnis: services/memory.py (SQLite/WAL, 5 Kategorien,
Dedupe-Kurator deterministisch), routers/memory.py (CRUD/export/dedupe).
MCP: mcp/mcp_memory.py (Guard-Beschreibungen gegen 14B-Loop) + mcp/mcp_mc.py
NEU (Stack-Management fuer Hermes: list/discover/register/route/restart/
status). Frontend MemoryView (Add/Filter/Suche/Delete/Aufraeumen).
Lokal verifiziert: CRUD + Dedupe (TestClient), MCP-Server syntax-OK,
Frontend-Build + Browser (MemoryView). Docs aktualisiert.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-25 08:16:12 +02:00
Hitonabi
6c8b6d81fe
feat(2.0): Phase 2 — System/OS + Connect
...
System: services/system.py (psutil CPU/RAM/Disk, sysfs GPU/Temp guarded),
routers/system.py GET status + restart (Whitelist) + self-update (systemd-
USER, sudo-frei). Connect: services/connect.py (Cline/OpenCode/Zed/Continue/
Claude Code/Memory-MCP → Gateway model:auto, LAN-IP-Override), routers/
connect.py. Frontend: SystemView (Metrik-Bars) + ConnectView (Tool-Tabs,
Copy, IP-Override).
Lokal verifiziert: Backend-Smoke + Frontend-Build + Browser (System-Bars,
Connect-Snippets). Docs aktualisiert (README/STATUS/Plan).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-25 08:08:43 +02:00
Hitonabi
cff3f0b1a8
feat(2.0): Phase 1 — Engine + Routing (Herzstueck)
...
Backend-Services: fit/caps/sources (portiert), discover (live HF + Fit +
Caps + ranked recommendation), llama-swap write/register + groups (Ko-
Residenz swap:false), LiteLLM-Gateway-Config + gateway-Service (model:auto +
Fallbacks). Router: discover/fit/register/groups/routing; health zeigt
gateway_reachable. Frontend: Modelle&Routing mit Caps-Chips, Fit-Badges,
Discover-Tab (live), Routing-View.
Lokal verifiziert: Backend-Smoke (alle Endpunkte) + Frontend-Build +
Browser (Shell, Discover, Caps/Fit). Box-Verifikation offen.
Docs: README + docs/STATUS.md (Phasen-Tracker + Resume-Guide).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-25 08:01:12 +02:00
Hitonabi
b805c294eb
feat(2.0): Phase 0 — Greenfield-Skeleton (FastAPI + React/shadcn)
...
Backend: FastAPI mit /api/health + /api/models (read-only aus llama-swap
config, Logik aus v1 portiert). Frontend: Vite + React + Tailwind v4 +
shadcn-Style App-Shell mit Cmd+K, Dark-default, PWA-Manifest; Modelle-View
live aus /api/models. Deploy-Geruest (systemd-Unit :9001, build/deploy-
Skripte, NOPASSWD-sudoers, kein Klartext-Passwort).
Verifiziert: Backend-Endpunkte + Frontend-Build + gerenderte Shell.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com >
2026-06-24 22:17:45 +02:00