How to import questions in bulk

What you'll need

• A SkillCort workspace with permission to create items
• Your existing questions in a text document, if you have them
• Optionally, a source document (syllabus, manual, policy) for AI grounding
• A rough idea of the folder structure your team will recognize

Step 1: Set up folders and folder-scoped tags first

Bulk import without structure just moves the mess into the platform. In the items library, create a folder per role or program, and define folder-scoped tags inside it — by competency, difficulty, or topic. Tags being scoped to the folder means the "escalation" tag in your support folder never collides with someone else's use of the same word elsewhere.

Decide the scheme before importing, because you will apply it as items land. A hundred items tagged on arrival take minutes; a hundred untagged items rediscovered next quarter take an afternoon. The picker modal you will use later filters by exactly these folders and tags, so this step is what makes future imports fast.

Step 2: Quick-paste your existing multiple-choice questions

If you have multiple-choice questions in a document, the quick-paste flow turns them into items without one-by-one entry. Paste the batch of questions with their options and correct answers, and SkillCort parses them into individual multiple-choice items in the current folder.

Skim the parsed results before moving on: check that options split correctly and the right answer is marked on each item, especially if your source formatting was inconsistent. Fixing a mis-parsed option now costs seconds; discovering it after candidates have answered costs a scoring correction.

Step 3: Generate items with AI, grounded in your material

For net-new content, use AI generation in the items library. State how many items you want — up to 100 per run — along with the topic and item type. For content that must match your world rather than general knowledge, upload a source document as grounding: a product manual, a syllabus, a policy PDF. The generator drafts items from that material.

Grounding is the difference between questions about your actual refund policy and questions about refund policies in general. Run several smaller, focused batches — one per competency or chapter — rather than one giant run; review stays manageable and tagging stays clean.

• State an explicit count, up to 100 items per run
• Upload a source document when accuracy to your material matters
• One focused batch per topic beats one sprawling batch
• Generated items land as drafts to review, not finished questions

Step 4: Review every generated draft before it counts

Treat AI-generated items as a first draft from a fast but unaccountable colleague. Review each one for factual accuracy, exactly one defensible correct answer, plausible distractors, and wording that tests the skill rather than reading comprehension of the question itself. Delete weak items without guilt — over-generating and cutting hard is the efficient pattern.

This review is where the human stays in charge of what candidates are measured on. An unreviewed generated item that reaches candidates is your assessment making a claim nobody checked. Tag the survivors as you review, so the pass leaves the library both vetted and organized.

Step 5: Import into an assessment through the picker modal

In the assessment builder, open the import picker to pull library items into your sections and pages. Filter by folder, tag, and item type to narrow hundreds of items down to the relevant dozen, then multi-select and add them in bulk. This is where the tagging discipline from step one pays off directly.

Combine filters to compose deliberately — for example, the support folder, tag "de-escalation", type multiple choice — instead of scrolling and cherry-picking. Composing from filters keeps the assessment aligned with the structure you planned, and makes it easy to explain later why these items are in this exam.

Step 6: Keep the library maintained as it grows

A bulk-fed library grows fast, so give it light routine care. Retire items that turn out to be ambiguous or leaked, keep tags meaningful as topics evolve, and add new generation runs into the existing folder scheme rather than inventing a parallel one per hire.

Because assessments import from the library, a well-kept library compounds: the next assessment for the same role is mostly a picker-modal session, not an authoring project. Shuffle and section weights then differentiate exams built from the same pool, which is exactly how parallel certification forms are maintained.