[workshop-sim] Improve: Add active-recall check for the two-file structure concept in Step 5

The file to edit is 05-agentic-workflows-intro.md. Read the existing content, especially lines 42–53, to understand the current explanation of the two-file structure (.md source vs .lock.yml compiled file). Design a short fill-in-the-blank or matching exercise that asks the learner to pair the file types with their purposes before the answer is revealed. Add this as a new, distinct callout block before the explanation, preserving all existing activities. Verify the change does not lower the documented quantitative scores.

Problem statement

05-agentic-workflows-intro.md has the lowest active_learning score in the entire 30-step corpus (2.4/10) and is the second-highest-dropout step in the Monte Carlo simulation: 18.6% conditional dropout among 37,457 at-risk runs (95% interval 18.2%–19.0%), with agentic-concept-gap as the sole failure category. Content inspection confirms the page does contain three "Activity" sections (predict-the-two-files, agentic-or-standard classification, write-a-task-brief), but they are placed after long blocks of prose explanation, and the core "two-file structure" concept (.md source vs .lock.yml compiled file) — the single idea Step 8 depends on most — gets only one short prediction prompt before the answer is revealed inline in the same paragraph (lines 42–53), rather than a standalone active-recall check.

Proposed change

Restructure the "two-file structure" section so the learner must commit to an answer (e.g., a short fill-in-the-blank or matching exercise pairing ".md" / ".lock.yml" with "you edit this" / "GitHub Actions runs this") in its own callout block before the explanation, mirroring the reveal-pattern already used successfully in "Activity 2" and "Activity 3" later on the same page. Do not remove or shorten the existing three activities — add one earlier, lower-stakes active-recall check for the two-file concept specifically, since it is the concept Step 8's dropout depends on most heavily.

Failure mode classification

Learning barrier (insufficient active-learning scaffolding for the highest-leverage concept on the page).

Quantitative guardrail

• Current overall_score for 05-agentic-workflows-intro.md: 5.43/10
• Weakest rubric dimension: active_learning at 2.4/10 (lowest in the corpus)
• Current learning KPI index for this step: (2.0×2.4 + 2.0×0.0 + 1.5×5.0) / 5.5 ≈ 2.24/10 (using current, pre-detection-fix checkpoint_quality of 0.0)
• Adding one more predict-then-reveal activity increases activity_density and the active-learning signal without adding conceptual load or removing any existing content, so overall_score and the learning KPI index should only rise, never fall.

Acceptance criteria

• The shared curriculum quantitative assessment overall_score for 05-agentic-workflows-intro.md stays flat or improves after the change
• The step's learning KPI index (2.0 × active_learning + 2.0 × checkpoint_quality + 1.5 × scaffolding) / 5.5 stays flat or improves, confirming students who remain in the workshop are still building skills effectively
• The three existing activities (two-file prediction, agentic-vs-standard classification, task-brief writing) and the existing checkpoint checklist are preserved and not weakened or removed
• The new active-recall check for the two-file structure is added as its own distinct exercise, not merged into the existing prose explanation

Suggested owner

workshop maintainer
Related to #3772

Generated by 🔬 Workshop Student Simulator · copilot · auto · 247 AIC · ⌖ 9.63 AIC · ⊞ 14.8K · ◷

• expires on Sep 25, 2026, 8:00 AM UTC