Originally created by: Akarsh-Hegde
Resolves all open CodeRabbit comments on PR [#347] (pre-main → main).
Changes
services/agents/_state.py
- Update wire-contract comment:
PrefetchStatus in mlx_server.rs already decodes speed; the comment was stale saying "exactly state/received/total/error".
services/agents/routes/prefetch.py (also has the _speed_state dict fix from the first commit)
log.error → log.exception in _run_prefetch failure handler — preserves the traceback that log.error dropped (Ruff TRY400; span.record_exception was already there, the log was the gap)
zip(strict=True) on specs/probe_results — documents and enforces the equal-length invariant guaranteed by asyncio.gather (Ruff B905)
services/agents/session_distiller.py
- Add
-> tuple return type to _get_embedder (Ruff ANN202)
- Split
import mlx.core as mx; mx.clear_cache() onto two lines (Ruff E702)
ui/app/setup/steps.tsx + page.tsx (🟠 Major fix)
- Dead-end in the Local-intelligence step: when the MLX runtime is unavailable (
runtime_found=false, runtime_installed=false, download_available=false) the unavailable state showed a message but no Retry, and canNext was s.modelReady — always false. User could never finish setup.
- Fix:
canNext now also gates open for the unavailable case, matching the same condition MLXBody uses to render the "not available for this Mac" message.
- Also correct two stale
// Step N comments in page.tsx that were left over from the Integrations/MLX tab reorder.
🤖 Generated with Claude Code
Originally posted by: coderabbitai[bot]
✨ Finishing Touches
🧪 Generate unit tests (beta)
- [ ] Create PR with unit tests - [ ] Commit unit tests in branch `fix/prefetch-speed-state-lint`Comment
@coderabbitai helpto get the list of available commands.Ticket changed by: Akarsh-Hegde