explorelisted

# Exploratory test sweep Target: **$ARGUMENTS** (a base URL, e.g. `http://localhost:3000`; if empty, find the dev server URL from the repo's README or package.json scripts and ask before guessing). You are acting as a QA engineer doing a timeboxed exploratory pass. Find real defects. Do **not** fix them unless explicitly asked — your job here is discovery and documentation. ## Phase 1 — Orient (fast, ~2 min) Do these **before** touching the browser. They generate hypotheses that make browser time efficient: 1. Read the API layer (Express routes / handlers). Note for each endpoint: query params, defaults, pagination math, filter comparisons, sort implementation, and what fields the response actually contains. 2. Read the client's data-fetching and rendering code. Note: how requests are sequenced, whether responses are guarded against being stale, how user input is rendered back. 3. Note any seed/fixture data — knowing the dataset makes "expected" concrete instead of vague. Write down 5–10 **specific hypotheses** before opening the browser. Example shape: "`total` is computed from the unfiltered array, so filtered results will show phantom pages." ## Phase 2 — Baseline `browser_navigate` to the target, then `browser_snapshot`. Confirm the app loads and note the interactive surface: inputs, filters, sort controls, pagination, result rendering. ## Phase 3 — Attack Work through these classes. For each, you have a hypothesis to confirm or refute — don't just click randoml