File: workshop/side-quest-11-06-anthropic-key.md
Overall Score: 5.39 / 10.0 (corpus mean: 6.14)
Flagged Dimensions:
| Dimension |
Score |
Benchmark |
Delta |
| cognitive_load |
~3.9 (concept component low; 26 new concepts vs ≤15 ideal) |
≤ 15 new concepts |
large |
| active_learning |
3.5 |
density ≥ 3 → 10 |
-6.5 |
Root Cause (≤ 2 sentences):
Despite a short 665-word body, the page introduces 26 distinct bolded/code terms (ANTHROPIC_API_KEY, engine: claude, api.anthropic.com, ANTHROPIC-API-KEY, CLAUDE_API_KEY, network.allowed, copilot-requests: write, plus every table cell and code-fence token), driven largely by an exhaustive "common mistakes" list of secret-name variants that adds vocabulary without adding a new procedural step. missing_checkpoint is again a false positive from the shortcode regex bug — the page has a real 5-item checkpoint at line 109.
Evidence (quoted from the file):
"Wrong name: any variation (anthropic_api_key, ANTHROPIC-API-KEY, CLAUDE_API_KEY) will cause a silent auth failure... Copied with extra whitespace... Closed the Anthropic tab before saving... Network allow-list missing..."
Learning Science Rationale:
Sweller's Cognitive Load Theory predicts that enumerating multiple near-identical failure variants (four bolded "common mistakes," each introducing its own bolded term) increases extraneous load without adding to the learner's actual procedural skill, since the underlying fix ("use the exact secret name") is a single idea repeated in four surface forms — this is the kind of redundancy the theory specifically flags as harmful to schema construction.
Improvement Prompt (for an agent):
Edit workshop/side-quest-11-06-anthropic-key.md to reduce concept density. Specifically:
1. Collapse the "Common mistakes with this secret" details block from four separately-bolded bullet items into a single short bulleted list under one heading term (e.g. "Common mistakes") without bolding each variant name as its own concept — keep the substance but stop treating each incorrect spelling as a distinct new term.
2. In the "Update your workflow frontmatter" code block, keep only the fields the learner will actually change (engine, network.allowed) and move the note about removing copilot-requests: write into a single inline sentence rather than a separate callout-style aside.
3. Consider adding one short hands-on check between "Get an Anthropic API key" and "Store the key as a repository secret" (e.g. a one-line command to confirm the key format starts with sk-ant-) to raise activity_density.
Re-run the corpus scoring script and confirm new_concepts drops toward 15-18 while checklist_items and activity_density are unchanged or improved.
Expected Score After Fix: 6.2 / 10.0
Generated by 🔬 Curriculum Quality Evaluator · copilot · auto · 76.7 AIC · ⌖ 17.4 AIC · ⊞ 9.5K · ◷
File:
workshop/side-quest-11-06-anthropic-key.mdOverall Score:
5.39 / 10.0(corpus mean:6.14)Flagged Dimensions:
Root Cause (≤ 2 sentences):
Despite a short 665-word body, the page introduces 26 distinct bolded/code terms (
ANTHROPIC_API_KEY,engine: claude,api.anthropic.com,ANTHROPIC-API-KEY,CLAUDE_API_KEY,network.allowed,copilot-requests: write, plus every table cell and code-fence token), driven largely by an exhaustive "common mistakes" list of secret-name variants that adds vocabulary without adding a new procedural step.missing_checkpointis again a false positive from the shortcode regex bug — the page has a real 5-item checkpoint at line 109.Evidence (quoted from the file):
Learning Science Rationale:
Sweller's Cognitive Load Theory predicts that enumerating multiple near-identical failure variants (four bolded "common mistakes," each introducing its own bolded term) increases extraneous load without adding to the learner's actual procedural skill, since the underlying fix ("use the exact secret name") is a single idea repeated in four surface forms — this is the kind of redundancy the theory specifically flags as harmful to schema construction.
Improvement Prompt (for an agent):
Expected Score After Fix:
6.2 / 10.0