Problem statement
04-github-actions-intro.md is the lowest-scoring curriculum step (overall score 5.39/10) and its dominant failure category is concept-overload (6,714 failures, 15.2% conditional dropout among 44,168 at-risk runs — the 3rd highest dropout step overall). The page introduces five new vocabulary terms (trigger, job, runner, step, action) plus a classic-vs-agentic comparison table in a single pass, then jumps straight to an unguided "label each part" exercise with only a collapsed answer key as a fallback (active_learning = 3.9/10, checkpoint_quality = 0.0/10 under the shared rubric).
Proposed change
Insert one fully-worked example between the "Label a sample workflow" exercise and the "Reveal the labels" answer key: show one line already labeled with a one-sentence rationale, then let learners label the remaining four lines themselves. This gives an intermediate scaffold instead of jumping from zero examples to a five-item unguided test.
Failure mode classification
Learning barrier — this is conceptual density and insufficient scaffolding before an unguided assessment, not an access or tooling problem (this step requires no terminal, credentials, or tooling per its own "Before You Start" section).
Quantitative guardrail
- Current
overall_score: 5.39/10 (lowest in the corpus)
- Weakest rubric dimensions:
checkpoint_quality = 0.0/10, active_learning = 3.9/10
- Current learning KPI index for this step:
(2.0×3.9 + 2.0×0.0 + 1.5×5.0)/5.5 = 2.7/10
- The proposed worked example directly targets
active_learning (adds a second guided practice rep) and should raise, not lower, scaffolding (currently 5.0/10, since has_prereq_section is already true) — it must not reduce the five-term vocabulary set or remove the unguided portion of the exercise, since that would lower cognitive challenge without addressing the actual gap (missing intermediate scaffold).
Acceptance criteria
Suggested owner
workshop maintainer (content addition with copilot coding agent assist for drafting the worked example)
Related to #3478
Generated by 🔬 Workshop Student Simulator · copilot · auto · 131.4 AIC · ⌖ 7.67 AIC · ⊞ 15.2K · ◷
Problem statement
04-github-actions-intro.mdis the lowest-scoring curriculum step (overall score 5.39/10) and its dominant failure category isconcept-overload(6,714 failures, 15.2% conditional dropout among 44,168 at-risk runs — the 3rd highest dropout step overall). The page introduces five new vocabulary terms (trigger, job, runner, step, action) plus a classic-vs-agentic comparison table in a single pass, then jumps straight to an unguided "label each part" exercise with only a collapsed answer key as a fallback (active_learning= 3.9/10,checkpoint_quality= 0.0/10 under the shared rubric).Proposed change
Insert one fully-worked example between the "Label a sample workflow" exercise and the "Reveal the labels" answer key: show one line already labeled with a one-sentence rationale, then let learners label the remaining four lines themselves. This gives an intermediate scaffold instead of jumping from zero examples to a five-item unguided test.
Failure mode classification
Learning barrier — this is conceptual density and insufficient scaffolding before an unguided assessment, not an access or tooling problem (this step requires no terminal, credentials, or tooling per its own "Before You Start" section).
Quantitative guardrail
overall_score: 5.39/10 (lowest in the corpus)checkpoint_quality= 0.0/10,active_learning= 3.9/10(2.0×3.9 + 2.0×0.0 + 1.5×5.0)/5.5 = 2.7/10active_learning(adds a second guided practice rep) and should raise, not lower,scaffolding(currently 5.0/10, sincehas_prereq_sectionis already true) — it must not reduce the five-term vocabulary set or remove the unguided portion of the exercise, since that would lower cognitive challenge without addressing the actual gap (missing intermediate scaffold).Acceptance criteria
overall_scorefor04-github-actions-intro.mdstays flat or improves after the change(2.0 × active_learning + 2.0 × checkpoint_quality + 1.5 × scaffolding) / 5.5stays flat or improvesconcept-overloadfailure count for04-actions-introSuggested owner
workshop maintainer(content addition withcopilot coding agentassist for drafting the worked example)Related to #3478