Issues

• Health check reports tiers 'healthy' while the vLLM engine loop crashes on every request

  weklund/mlx-stack#52 · Filter by author weklund opened on Jun 17, 2026

• Default continuous_batching breaks inference: ArraysCache.__init__() missing 'size' (vllm-mlx 0.2.6)

  weklund/mlx-stack#51 · Filter by author weklund opened on Jun 17, 2026

• up: models already in the HuggingFace cache aren't detected (forces redundant re-download)

  weklund/mlx-stack#50 · Filter by author weklund opened on Jun 17, 2026

• Add mlx-stack integrate command for AI agent harness onboarding

  weklund/mlx-stack#49 · Filter by author weklund-agent opened on Apr 7, 2026

• Gemma 4 blocked: mlx_lm 0.31.1 lacks gemma4 architecture module

  weklund/mlx-stack#48 · Filter by author weklund-agent opened on Apr 7, 2026

• Add Gemma 4, LFM2, and Qwen3-Coder models to catalog

  weklund/mlx-stack#47 · Filter by author weklund-agent opened on Apr 7, 2026

• feat: auto-upgrade dependency tools (vllm-mlx, litellm) on version mismatch

  enhancement
  weklund/mlx-stack#46 · Filter by author weklund opened on Apr 5, 2026

• bug: config set treats negative numeric values as CLI flags

  bug
  weklund/mlx-stack#45 · Filter by author weklund opened on Apr 5, 2026

• bug: bench leaves stale PID file when temporary instance fails to start

  bug
  weklund/mlx-stack#44 · Filter by author weklund opened on Apr 5, 2026

• bug: setup interactive model selection crashes on invalid input

  bug
  weklund/mlx-stack#43 · Filter by author weklund opened on Apr 5, 2026

• Rework tier system: decouple hardware-aware recommendation from rigid tier assignment

  weklund/mlx-stack#39 · Filter by author weklund opened on Apr 4, 2026
  1

• Multi-Mac mesh networking: transparent distributed serving across Apple Silicon machines

  weklund/mlx-stack#31 · Filter by author weklund-agent opened on Apr 4, 2026

• Agent-aware prefix cache sharing for KV cache reuse across requests

  weklund/mlx-stack#30 · Filter by author weklund-agent opened on Apr 4, 2026
  1

• Agent reliability layer: circuit breakers, loop detection, and automatic model escalation

  weklund/mlx-stack#29 · Filter by author weklund-agent opened on Apr 4, 2026
  2

• Embeddings endpoint for local RAG and semantic search

  weklund/mlx-stack#28 · Filter by author weklund-agent opened on Apr 4, 2026
  3

• Agent harness integration guides and config generators

  weklund/mlx-stack#26 · Filter by author weklund-agent opened on Apr 4, 2026

• Smoke test ReadTimeout: health check passes but inference times out for qwen3.5-0.8b

  weklund/mlx-stack#21 · Filter by author weklund opened on Apr 3, 2026

• Integration test flaky: port still bound after shutdown in test_full_lifecycle

  weklund/mlx-stack#20 · Filter by author weklund opened on Apr 3, 2026

• Run integration tests in CI on Apple Silicon runners

  weklund/mlx-stack#18 · Filter by author weklund opened on Apr 3, 2026

• vllm-mlx continuous batching disabled due to upstream bug (waybarrios/vllm-mlx#211)

  weklund/mlx-stack#17 · Filter by author weklund opened on Apr 3, 2026

• Qwen 3.5 0.8B fails with ArraysCache error on vllm-mlx — fast tier unusable

  weklund/mlx-stack#15 · Filter by author weklund-agent opened on Apr 2, 2026