⚠️They are ordered by relevance according to @SEgt-WMF , but will be reviewed by @Isaac as well 😊

• Pageviews column treats large numbers as decimals when sorting

When sorting the pageviews column numerically, values like "100333" are being interpreted as "100.333" due to comma/thousands separator parsing issues. This results in an incorrect sort order, with smaller numbers appearing before larger ones.
Expected: Pageview numbers should be sorted as integers, with proper handling of thousands separators. "100,333" should sort higher than "50,000" and lower than "200,000".

• Replace standalone quality column with integrated quality progress bar

The current "Quality" column occupies space and duplicates information already shown in the progress bar.
Expected: Remove the separate Quality column and incorporate the term quality directly into the progress bar column (e.g., "Quality progress: 65%"). This simplifies the interface and reduces visual clutter.

• Clarify "Add References" vs. "Add Sources" in micro-tasks

The micro-task generator offers two seemingly similar task types: "Add References" and "Add Sources." This distinction is unclear to users.
Expected: Provide tooltips or inline documentation explaining the difference (e.g., "Add References" = citation needed tags; "Add Sources" = articles lacking any sources). Alternatively, consider merging or renaming for clarity.

• Remove regions from "topics" and just leave regional selectors as part of countries.

Expected: Any topic from the articletopic model that starts with "Geography" is excluded.

• Fetch pageview counts even when titles need to be normalized. Right now, if the first letter in the title is lowercase, features like topics/quality still work but pageviews return zero. Note: this seems to be relevant to language link counts in the current MR too so will need fixed there too.

Expected: Use the output of the revision API call to set the canonical name for the article and use that for any future API calls. If this means that we also only use the canonical titles in the output table (post normalization/redirects) and not the specific titles input by the user, I think that's okay.

• Make results table sortable

The output table is currently static. Columns (e.g., article title, quality signal, namespace, etc.) should be sortable so users can prioritize tasks more easily.

• Incorrect status shown for non-existing articles

When an article in the input list does not exist on the target wiki, the tool sometimes shows “Up to date”, which is misleading.
Expected: Non-existing pages (e.g., Telugu: మాక్స్ వెబర్) should be clearly flagged as *“Page does not exist”* (or similar), not treated as valid articles.

• Namespace handling bug with category results that include Talk pages

When categories return pages from Namespace 1 (Talk:), the tool should remove everything before the colon to get the correct names of articles
Expected: Talk:ArticleName becomes ArticleName.

• Duplicate recommendations in output

The same article can appear multiple times in the recommendation list.
Expected: Each article should appear only once per generated task list.

• Don't split page titles on commas

Commas are acceptable values in page titles. For example, if I wanted data on en:Paris, Texas, the tool actually splits this into en:Paris and en:Texas. Newline should be the only split character in that input method I think.

• Unhelpful error messages

Errors such as “Error fetching article quality” do not guide the user.
Expected: Provide actionable guidance, e.g., suggest reducing batch size, retrying after a short wait, or indicating possible API latency/timeout (may be related to existing latency ticket).

• No loading indicators for long operations

When processing large lists, the interface appears unresponsive, which may lead users to think the tool has frozen.
Expected: Show a loading indicator or progress state during long-running requests.