Walkthrough: one course, step by step

Excerpts from the records of a pilot course, October 2026

Curriculus is a free set of instructions and checking scripts that a college's own staff run with AI assistants to build a whole course, from the syllabus to the slides. The How it works page explains what Curriculus is and who does what.

Curriculus keeps a record of every check, finding, fix and approval while it builds a course. This page shows short excerpts from the records of a pilot course, Introduction to General Psychology, so you can see the work behind each step.

We copied each excerpt word for word. We removed the names of people and of the school, shortened some file names and marked each cut with [...]. The procedure's own instructions, scripts and templates don't appear on this page, because member schools receive them with their copy of Curriculus.

A few terms help. A "script" is a small program that checks the work automatically, and "FAIL" and "WARN" are the problems a script reports. A "session" is one run of an AI assistant. "U04" means Unit 4, which teaches classical and operant conditioning. A reviewer marks each finding "Must" or "Should," and a Must finding needs a fix or a ruling from the Lead before the step can close. We replaced the name of the main textbook's author with "Author."

Step 0: New course

Step 0 sets up the course folder, copies in each file the instructor sent and records the scope that the Lead set for the course. Curriculus builds from the copies and never edits the instructor's originals.

The course folder, two levels deep. The instructor's files, the working drafts, the approved files and the record each have their own folder.

00_Intake/     activity_assets  correspondence  decks  readings  syllabus_and_planning  kickoff_notes.md
10_Working/    step1  step2  step4  step5  step6  step7  _cleared
20_Final/      00_Course_Navigator  10_Course_Overview  20_Detailed_Lessons  30_Assessments
30_Record/     assessments  archive  course_history  delete_lists  lead_replies  lessons  qa  reading_analysis  structure
COURSE_LOG.md  course.json  course_record.json

From a listing of the course folder

The scope that the Lead set for this course.

Scope:
- 14 teaching units across 16 learning weeks. Leave out break-only and exam-only weeks.
- Call each teaching block a unit.
- Graded work is out of scope: no rubrics, points or grading sections.
- The 2017 syllabus describes a past offering. Don't carry its dates, room, meeting times or instructor into this course as current facts.

From 00_Intake/kickoff_notes.md

Step 1: Course overview

Step 1 turns the syllabus and the course requirements into the course overview. An automatic check confirms that the overview matches the course record word for word, a second AI checks the draft, and the Lead approves the final version.

The review of the first draft found 13 problems. The AI that made the fixes fixed 12 of the 13 and gave its reason for leaving one finding unchanged.

The automatic check on the approved overview.

Run: 2026-10-08 13:53

Validator, gate: step6
Result: 0 FAIL, 0 TODO, 0 WARN. Ready for Step 6.

From 30_Record/qa/step1_qa.md

One reviewer finding on the first draft, and the fix.

#	Location	What's wrong	Why it matters	Must or should	Suggested fix
9	Instructor Appendix, U14.LO2; `course.json` `units[U14].objectives[U14.LO2].text`	"the major kinds of struggle that Chapter 16 presents" does not name the text, and the course assigns two texts.	The objective is read on its own in later steps.	Should	Write "that the author's Chapter 16 presents."

#	Result	What changed
9	Fixed	`U14.LO2` reads "the author's Chapter 16".

From 30_Record/qa/step1_qa.md

The Lead's reply, saved word for word, and the line Curriculus added to the course log.

All looks good.

- October 8, 2026: Step 1 course overview approved. Course_Overview.docx, document version 9, SHA-256 1595ab77f60c9a8f. QA report: 30_Record/qa/step1_qa.md. [...]

From 30_Record/lead_replies/step1 and COURSE_LOG.md

Step 2: Reading analysis

Step 2 writes a digest of each reading, with a page number for every point. For each page number, the digest copies a short phrase from that page, and an automatic check confirms that the phrase really appears there. A second AI then checks each digest against its reading.

The closing report for Step 2.

Run: 2026-10-07 13:39

- Assignments in the plan: 18. Analyzed: 16. Reused: 0. Without a copy: 2 (U07_R01, U09_R01).
- Script checks: 0 FAIL, 0 WARN (0 with a reason).
- Review findings: 7 Must, 35 Should. Fixes rows: 42 fixed, 0 left unchanged. Must findings fixed: 7 of 7.

From 30_Record/qa/step2_qa.md

The automatic check on page numbers, after every fix.

[...] The script found all 399 anchors on their pages, the 13 new ones among them.

From 30_Record/qa/step2_qa.md

A reviewer finding on the Unit 4 reading, and the fix.

#	Entry and item	What's wrong	The reading says	Must or should	Suggested fix
4	U04_R01, theme 2, "Classical conditioning: learning by pairing"	The locators are pp. 9 to 18 and p. 15. The description's example of "the smell of a grandmother's cookies" is on p. 4, in the author's introduction, and not within the locators.	p. 4: "like a bird singing or a firecracker exploding or the smell of our grandmother's cookies, we pair it lawfully with something happening inside of us".	Should	Add a p. 4 locator for the cookies example, or drop that example from the description.

#	Result	Note
4	fixed	`U04_R01`, theme 2: a p. 4 locator was added for the cookies example.

From 30_Record/qa/step2_qa.md

Step 3: Reading map

Step 3 builds one map of which readings belong to each unit and each week. Software builds the map from the course record, so the map can't disagree with the course record. The map also flags each reading that the course has no copy of.

Unit 4's entry in the reading map.

    {
      "unit_id": "U04",
      "title": "Theory Story #2: “Maybe it’s really all about what happens outside of us…” Behavioral Bases of Experience",
      "week_ids": [
        "W04",
        "W05"
      ],
      "resource_ids": [
        "U04_R01"
      ],
      "digest": "digests/readings_digest_U04.json"
    },

From 30_Record/reading_analysis/reading_map.json

The course log line.

- October 7, 2026: Step 3 reading map built. reading_map.json, SHA-256 c0a3e93f258bccfd.

From COURSE_LOG.md

Step 4: Curriculum structure

Step 4 places each unit in one of five Parts of the course and names the units it builds on. An automatic check tests the placements, a second AI checks them, and the Lead approves them.

The automatic check on the placements the Lead approved.

Checked placements_v5.json.

Learning domain	Part 1: Foundation	Part 2: Building	Part 3: Breakthrough	Part 4: Advanced
D2	U01	No unit.	No unit.	U08; U09; U10; U11; U12
D1	U02	U03; U04; U05	U06; U07	No unit.

Part 5: Proficiency, shared by all the domains: U13 (D2); U14 (D2).

Gaps: none. Coverage is complete.

Totals: 14 placements, 0 gaps, 0 FAIL, 0 WARN (0 with a reason).

From 30_Record/qa/step4_qa.md

A reviewer finding on Unit 4, and the fix.

#	Unit	What's wrong	The course says	Must or should	Suggested fix
4	U04	The reason says U05 widens the comparison to three accounts. U05.LO4 compares two, and the reason for U05 says so as well: "the biological story of U03 or the behavioral story of U04". The U05 outcome does say "the biological and behavioral accounts", but the lesson is written from the objective, and the instructor would read two different descriptions of U05 on the pathway.	U05.LO4: "Compare the psychoanalytic account of an experience with a biological or behavioral account, naming one assumption of each."	Should	Replace the last sentence with: "Comparing accounts of the same behavior is the work that U05, on the psychoanalytic story, continues with a third story, and that U06, on the humanistic, existential and transpersonal story, and U07, on culture and context, widen to four and five accounts."

#	Result	Note
4	fixed	The last sentence of U04's reason now says U05 continues the comparison with a third story, in the wording the review suggested.

From 30_Record/qa/step4_qa.md

The course log line.

- October 8, 2026: Step 4 curriculum structure approved. placements_v5.json, 14 units placed, no gaps, course.json SHA-256 46fd5906ec1c963d. QA report: 30_Record/qa/step4_qa.md

From COURSE_LOG.md

Step 5: Course Navigator

Step 5 builds one web page where the instructor sees each unit's place in the course and confirms each placement. The instructor opens the page with a double-click and needs no account.

The note the page showed the instructor about two readings the course had no copy of.

U07. We have no copy of Chapter 7, Culture-Context, so this unit was placed without it. Its placement may change once we have the reading. U09. We have no copy of Chapter 9, Consciousness, so this unit was placed without it. Its placement may change once we have the reading.

From the Course Navigator page

The first version of the page, with 14 placements waiting on the instructor, and the current version, with none waiting.

- October 8, 2026: Step 5 Course Navigator approved. Course_Navigator.html, version 1, 14 units on the pathway, 14 placements waiting on the instructor, page SHA-256 0fedccf248e63141.

- October 11, 2026: Step 5 Course Navigator approved. Course_Navigator.html, version 8, 14 units on the pathway, 0 placements waiting on the instructor, page SHA-256 6eb80ae0d103fdf2. [...]

From COURSE_LOG.md

The Lead's reply after the instructor's review, saved word for word.

Confirm all 14 placements. We can hide the Notes from Curriculus note.

From 30_Record/lead_replies/step5

Step 6: Lesson guides

Step 6 writes one lesson guide for each unit. An automatic check compares every citation with the readings and their pages, and every quotation of eight words or more with the page it cites. Two more AI reviewers then check the guide, starting fresh: one against the readings and one as a lesson.

The first review of the Unit 4 guide found 43 problems, and the AI that made the fixes fixed all 43.

The automatic check and the Word file build for the first approved version of the Unit 4 guide.

Run: 2026-10-09 09:05

Checked 10_Working/step6/U04/Lesson_Guide_U04_v2.md.
Totals: 0 FAIL, 0 WARN (0 with a reason).
No findings.

Word build:
Built: 10_Working/step6/U04/Lesson_Guide_U04_v2.docx
Quick checks: passed (121 list items, 5 tables, 14 links, 49 headings, 0 figures).
Properties: Document version 2. Instruction version Step 6 v4.0. Build date 2026-10-09.
Pipeline-word scan of the Word text: 0 to fix, 1 to check.
Render check: 33 pages. Instructor Notes starts page 24. Notes from Curriculus starts page 33.

From the closed records of the Unit 4 guide

A reviewer finding: the guide had dropped the author's hedge. The fix puts the hedge back.

#	Section	What's wrong	The text	Must or should	Suggested fix
16	9. Practice Variations & Extended Learning, and 5. The Core Concept	The author's hedge is gone. He says effective punishment tends to happen under these conditions.	Guide, Section 9: "It then lists what makes a punishing consequence effective". Guide, Section 5: "he lists six conditions under which a punishing consequence is effective". Reading, p. 24: "Effective punishment tends to happen when the consequence:"	Should	In Section 9 write "It then lists when a punishing consequence tends to be effective," and "According to the list, a punishing consequence tends to be effective when it:". In Section 5 write "six conditions under which a punishing consequence tends to be effective."

Finding	Review	Result	What changed and where, or the text that shows the finding is wrong
16	source	fixed	9. Practice Variations & Extended Learning, "Measure a punishment against the list": now "It then lists when a punishing consequence tends to be effective" and "a punishing consequence tends to be effective when it:". 5. The Core Concept, "Consequences": now "six conditions under which a punishing consequence tends to be effective".

From the closed records of the Unit 4 guide

Part of the figure check. Every figure in the chapter is either used or left out with a reason.

Item	Reading and page	In the guide	Reason when left out
Figure 4.4, a dog before, during and after conditioning	Author (2017, Ch. 4, p. 10)	Not in the guide. The sequence is a numbered list in 5. The Core Concept	The instructor corrects what the figure shows: it draws the neutral stimulus as a bell, and the author writes that Pavlov used a metronome (p. 8).
Figure 4.5, higher-order conditioning	Author (2017, Ch. 4, p. 11)	Not in the guide	The guide doesn't teach higher-order conditioning.
Figure 4.6, photo of a tourist with a stingray	Author (2017, Ch. 4, p. 12)	Not in the guide	The guide doesn't teach the Stingray City example.
Figure 4.7, curve of acquisition, extinction and spontaneous recovery	Author (2017, Ch. 4, p. 14)	5. The Core Concept, reproduced as figures/U04_fig1.png, directly under the four features of pairing

From 30_Record/qa/step6_qa.md

The Lead's approval, saved word for word, and the course log line.

I approve U04. I want to keep this lesson. Yes, the key facts are long, but it's still shorter than what it used to be. And we can always wait for more specific feedback about further reducing the length if needed.

- October 15, 2026: Step 6 lesson guide for U04 approved. Lesson_Guide_U04.docx, document version 8, SHA-256 8f7b98358215cba8. A polish pass of document version 7: one copy-edit session, no Reviewer, and a script confirmed that no citation, quotation, number, name, hedge or objective changed. [...]

From 30_Record/lead_replies/step6 and COURSE_LOG.md

Step 7: Assessments

Step 7 writes twenty questions for each unit, with answers and guidance for the instructor. An automatic check tests the questions, their citations and their quotations, and fresh AI reviewers check every answer against the reading.

The status table the Lead read before approving the Unit 4 assessment.

Unit	Draft	FAIL	WARN	Source review	Teaching review	Fixes	Word file	Ready for the Lead
U04	v3	0	0	v1: 3 Must, 12 Should	v1: 1 Must, 14 Should	30 of 30 answered, polished	built from v3	Ready for the Lead

From 30_Record/qa/step7_qa.md

Every one of the unit's objectives is tested by at least three questions.

Objective	Support in the guide	Tested by	Gap
Describe learning as the behavioral Theory Story defines it	supported	Q1, Q4, Q10, Q13, Q20	None
Explain classical conditioning, operant conditioning and observational learning	supported	Q2, Q7, Q12, Q14, Q16	None
Apply reinforcement to explain an everyday behavior	supported	Q3, Q5, Q8, Q9, Q11, Q13, Q15, Q18, Q19, Q20	None
Compare how the biological and behavioral Theory Stories would explain the same example	supported	Q17, Q19, Q20	None

From 30_Record/qa/step7_qa.md

A reviewer finding: an answer guide had dropped the author's hedge. The fix restores it.

#	Question	What's wrong	The text	Must or should	Suggested fix
11	Q16	The fourth indicator drops the author's hedge. Q14 states the same point as he does.	Indicator: "it says pairings tend to last for years (Author, 2017, Ch. 4, p. 15)." Reading: "these powerful pairings tend to hang around, sometimes for many years even without new experiences" (Author, 2017, Ch. 4, p. 15).	Should	Write: "it says pairings tend to hang around, sometimes for many years."

Finding	Review	Result	What changed and where, or the text that shows the finding is wrong
11	source	fixed	Q16, the fourth Response Indicator: "pairings tend to last for years" now reads "pairings tend to last, sometimes for many years."

From 30_Record/qa/step7_qa.md

The course log line.

- October 11, 2026: Step 7 assessment for U04 approved. Assessment_U04.docx, document version 3, SHA-256 efe20ca9349de803. [...]

From COURSE_LOG.md

Step 8: Presenter decks

Step 8 writes a PowerPoint deck for each unit and week, with speaker notes. Software builds the slides and checks every quotation of eight words or more against its page, and AI reviewers check the deck against the readings and as a lesson.

These excerpts come from a trial run of the Unit 4, Week 4 deck. The deck reached the Lead for reading, and the record holds no Lead approval yet.

The status table for the deck.

Deck	Deck file	Slides	FAIL	WARN	Source review	Teaching review	Fixes	PowerPoint file	Slide pictures	Ready for the Lead
U04_W04	v3	28	0	1 (1 with a reason)	v1: 3 Must, 8 Should	v1: 0 Must, 14 Should	25 of 25 answered, polished	built from v3 for the Lead	drawn	Ready for the Lead

From 30_Record/qa/step8_qa.md

Part of the quotation check. Each quotation on a slide matches the reading word for word, with its page.

Slide	Kind	Opens with	On the face	Cuts or reason
5	Quotation	merely and wonderfully, behavior	word for word	Author (2017, Ch. 4, p. 2).
6	Quotation	are not destinies	word for word	Author (2017, Ch. 4, p. 4).
7	Quotation	a relatively permanent change in behavior or knowledge	word for word	Author (2017, Ch. 4, p. 7).
8	Quotation	But he didn’t use a bell! He used a metronome	word for word	Author (2017, Ch. 4, p. 8).
11	Quotation	would be considered unethical by today’s standards	word for word	Author (2017, Ch. 4, p. 17).
13	Quotation	occurs immediately before the response	word for word	Two phrases from Table 4.1, Author (2017, Ch. 4, p. 18).
14	Table from the reading	Something is added to increase the likelihood of a behavior	word for word	Table 4.2, all rows, Author (2017, Ch. 4, pp. 19 to 20). The chapter's italics are bold on the face.
15	Quotation	anything that increases behavior is a reinforcer	word for word	Author (2017, Ch. 4, p. 21). In bold: the phrase the guide marks as the point.
16	Quotation	anything that reduces behavior is a punishment	word for word	Author (2017, Ch. 4, p. 23). Also "doesn’t punish speeding", Author (2017, Ch. 4, p. 24).
17	Quotation	the most powerful partial reinforcement schedule	word for word	Author (2017, Ch. 4, p. 31).

From 30_Record/qa/step8_qa.md

A reviewer finding: a slide had changed the author's question. The fix restores the question.

#	Slide	What's wrong	The text	Must or should	Suggested fix
6	10, face	The face changes the author's question. He asks about a date when you don't know how it will go, and the face makes it a date that might go badly, which hands the room the premise of the expected answer.	Deck: "The author asks which shirt to wear on a first date that might go badly." Reading, pp. 17 to 18: "Which shirt do you want to wear on a first date with someone, when you don’t know if it will go well or not?" Deck notes: "If the date goes badly, the favorite shirt may come to call that evening up."	Must	"The author asks which shirt to wear on a first date when you don’t know if it will go well or not. His hint: not your favorite (Author, 2017, Ch. 4, pp. 17 to 18). Why not?"

#	Review	Slide	Finding	Result	What changed
6	source	10, face	The face changes the author's question into a date that might go badly.	fixed	The face no longer says the date might go badly. Its preamble reads "You don’t know how a first date will go. The author’s hint: not your favorite shirt." and its question reads "Why not your favorite shirt?" A preamble's box holds one line, and the author's own wording with the hint needs two, so his wording, "when you don’t know if it will go well or not", is in the notes under "Activity", with its pages (pp. 17 to 18). The expected response's last sentence, that the hint depends on not knowing how the evening will go, now matches the face.

From 30_Record/qa/step8_qa.md

Get the full procedure

The How it works page shows four sample pages from this course. The slide comes from a trial deck that has no Lead approval yet, as Step 8 shows. Member schools receive the full procedure, with its instructions, scripts and templates.

Get started for free