[curriculum-eval] 04-github-actions-intro.md: cognitive_load — 28 new concepts nearly double the ideal for a 5-minute refresher

**File:** `workshop/04-github-actions-intro.md`
**Overall Score:** `5.39 / 10.0` (corpus mean: `6.14`)

**Flagged Dimensions:**

| Dimension | Score | Benchmark | Delta |
|---|---|---|---|
| active_learning | 3.9 | ≥3 activity density | -0.9 (density 1.19 vs 3.0 ideal) |
| cognitive_load | 6.8 | 10 (≤800 words, ≤15 concepts) | -3.2 (756 words is fine; 28 new concepts is the driver) |
| checkpoint_quality | 0.0 | 10 (≥4 checklist items) | -10.0 (scorer defect — see Root Cause) |

**Flagged issue `concept_overload` confirmed:** 28 new concepts (trigger, job, step, runner, action, workflow file, `on:`, `jobs:`, `steps:`, `runs-on:`, `uses:`, `run:`, self-hosted runner, agentic workflow, task brief, safe outputs, permissions model, etc.) is nearly double the 15-concept ideal, even though word count (756) is within budget.

**Root Cause (≤ 2 sentences):**
This is the workshop's first Actions primer, so it introduces the entire Actions vocabulary (trigger/job/step/runner/action) plus a preview comparison table against agentic workflows in one page, giving 28 distinct new terms — almost double the ideal — even though the prose itself is appropriately short. `checkpoint_quality` also shows 0.0 due to the corpus-wide `## :white_check_mark: Checkpoint` vs. `## ✅ Checkpoint` scorer mismatch; the real checkpoint has 5 concrete items and is not the actual gap.

**Evidence (quoted from the file):**
> A GitHub Actions workflow is a YAML file in `.github/workflows/` that tells GitHub:
> - _when_ to run (`on`)
> - _what_ to run (`jobs`)
> - _how_ each job executes (`steps`)

**Learning Science Rationale:**
Cognitive Load Theory (Sweller) distinguishes intrinsic load (the inherent difficulty of Actions vocabulary) from extraneous load (how it's presented); here the "Why This Matters for Agentic Workflows" comparison table adds four more concepts (permissions model, safe outputs, task brief execution) on top of the five core Actions primitives already taught in the same page, front-loading more novel terms than working memory can consolidate in one pass — especially for a page positioned as a "5 Minutes" refresher.

**Improvement Prompt (for an agent):**
```
Open workshop/04-github-actions-intro.md. Move the "Why This Matters for Agentic Workflows" comparison table (Trigger/Runner/Steps/Output classic-vs-agentic) out of this page and into the opening of workshop/05-agentic-workflows-intro.md instead, where those agentic-specific concepts (task brief, safe outputs, permissions model) are the primary learning objective and already have supporting context. Keep 04-github-actions-intro.md focused only on the five classic Actions primitives (trigger, job, runner, step, action) plus the existing labeling exercise and real-workflow exploration activity, reducing new_concepts from 28 toward 15-18. Update the transition sentence at the end of 04 so it still previews the connection to agentic workflows in one sentence without introducing new vocabulary. Run `npx --yes markdownlint-cli2 "**/*.md"` after editing and fix any lint errors.
```

**Expected Score After Fix:** `6.4 / 10.0`

> Generated by [🔬 Curriculum Quality Evaluator](https://github.com/githubnext/gh-aw-workshop/actions/runs/35026063971) · copilot · auto · 76.1 AIC · ⌖ 8.13 AIC · ⊞ 9.5K · [◷](https://github.com/search?q=repo%3Agithubnext%2Fgh-aw-workshop+is%3Aissue+%22gh-aw-workflow-call-id%3A+githubnext%2Fgh-aw-workshop%2Fcurriculum-evaluator%22&type=issues)
> - [x] expires on Sep 16, 2026, 9:39 PM UTC

Open workshop/04-github-actions-intro.md and workshop/05-agentic-workflows-intro.md. Move the comparison table as requested, keep the existing activities and a brief transition in 04, then run npx --yes markdownlint-cli2 "**/*.md". Done means 04 focuses on the classic Actions concepts, 05 contains the agentic comparison, and Markdown lint passes.

Written by the indexing model from the issue text.