Code

History

Commits on loop/professionalize to controller/src/modules35+
  1. Merge remote-tracking branch 'origin/main' into loop/professionalizeseroxdesign committed 3 months ago1e878d6
  2. fix(controller): parse the CUDA UMD version banner from newer drivers (#185)luminary19 committed 3 months ago56f691f
  3. refactor(controller): split engines god-router into concern registrarsseroxdesign committed 3 months ago232e97b
  4. refactor(controller): decompose the chat-completions handler into named stagesseroxdesign committed 3 months ago252add4
  5. refactor(controller): move engine-specific helpers out of backend-builder into their specsseroxdesign committed 3 months agod350a86
  6. refactor(controller): one process-inventory parser behind the ProcessRunner seamseroxdesign committed 3 months ago02c4dcf
  7. refactor(controller): collapse per-backend upgrade routes into /runtime/:backend/upgradeseroxdesign committed 3 months ago054c48b
  8. fix: keep long agent tool streams aliveseroxdesign committed 3 months agoe956b02
  9. refactor: share extra argument serialization (#179)borkiss committed 3 months agob7677d6
  10. fix(security): route homelab clients through tailnetseroxdesign committed 4 months ago6d56eb9
  11. perf: make usage analytics load immediatelyseroxdesign committed 4 months agoe1b3d85
  12. feat: manage isolated chatterbox voiceseroxdesign committed 4 months ago8a252d5
  13. fix: fail closed on gpu workload overlapseroxdesign committed 4 months agofa8ea94
  14. fix: quote docker multi-gpu selectorsseroxdesign committed 4 months ago35b8c6a
  15. fix: enforce isolated gpu ownershipseroxdesign committed 4 months agode84c7d
  16. feat: guard model launches by gpu leaseseroxdesign committed 4 months ago0fb1c1c
  17. feat: validate voice reference audioseroxdesign committed 4 months agoeb507ca
  18. feat: isolate gpu capability leasesseroxdesign committed 4 months ago2c4149b
  19. feat: encrypt local voice profilesseroxdesign committed 4 months agoad3402b
  20. feat: expose stable gpu identitiesseroxdesign committed 4 months ago6d6e001
  21. fix: keep recipe status out of engine argsseroxdesign committed 4 months ago3c7da56
  22. feat: configure explicit vision inputseroxdesign committed 4 months agoebee81a
  23. refactor: launch vllm from serve runtimeseroxdesign committed 4 months ago425cf03
  24. refactor: make serve runtime explicitseroxdesign committed 4 months ago97d18b5
  25. fix: replace loose null checks in rig accelerator parsingseroxdesign committed 4 months agobc356dd
  26. feat: add rig configuration resource with auto-detected local nodeseroxdesign committed 4 months ago42e13f5
  27. feat(perf): site-wide speed and simplification sweepseroxdesign committed 4 months ago76ce741
  28. feat(desktop): deploy a controller to any ssh host from the appseroxdesign committed 4 months ago0f9e97e
  29. fix(engines): point cmake at nvcc via CUDACXX for service-launched controllersseroxdesign committed 4 months agoeb7058d
  30. feat(engines): managed llama.cpp source build + wizard auto-installseroxdesign committed 4 months agoe0b45d1
  31. feat(onboarding): 3 starter presets + first-run redirectseroxdesign committed 4 months ago24c0e91
  32. fix: i20 verified findings — setup wizard, search races, recipe ids, dead codeseroxdesign committed 4 months agoaef3f98
  33. test(harness): injectable process + fetch seams; controller spawn paths testedseroxdesign committed 4 months ago6c35dec
  34. refactor(workspace): shared/ becomes @local-studio/contracts packageseroxdesign committed 4 months agof0c5a59
  35. fix(controller): running-model mis-ID, rate-limit IP spoofing, config truncation, recipe rangesseroxdesign committed 4 months ago2a90ab1