# Download Runner And Dashboard Fixes Implementation Plan

> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.

**Goal:** Fix catalog-sync so download jobs start promptly on NAS and dashboard download speed only reflects truly active download workers.

**Architecture:** Tighten dashboard worker selection to current running items only, and remove the pre-worker playlist export refresh that can block a download stage before any download worker starts. Keep playlist export behavior during item completion and stage finalization.

**Tech Stack:** Python, sqlite3, unittest, FastAPI ops dashboard

---

### Task 1: Lock The Regression With Tests

**Files:**
- Modify: `tests/catalogsync/test_ops_api.py`
- Modify: `tests/catalogsync/test_ops_runner.py`

- [ ] Add a dashboard API regression test that seeds one real running download worker plus stale historical workers and expects transfer speed to only include the live worker.
- [ ] Add a runner regression test that keeps a stage open with pending downloads and expects already-completed playlist exports not to run before pending download workers start.
- [ ] Run targeted tests first and confirm they fail for the expected reason.

### Task 2: Apply The Minimal Fix

**Files:**
- Modify: `musicdl/catalogsync/ops/web.py`
- Modify: `musicdl/catalogsync/ops/runner.py`

- [ ] Restrict dashboard worker rows to workers whose current job item is still running under an active job.
- [ ] Remove the pre-worker playlist artifact refresh from download stage startup so worker claiming is not blocked by export work.
- [ ] Keep existing per-playlist export on item completion and full export refresh on stage completion.

### Task 3: Verify The Fix

**Files:**
- Modify: `tests/catalogsync/test_ops_api.py`
- Modify: `tests/catalogsync/test_ops_runner.py`

- [ ] Run the focused regression tests and confirm they pass.
- [ ] Run a slightly wider ops test slice to catch nearby regressions.