Home / Workshop, capstone and exam
The capstone, three real projects to choose from
Shipping a small real feature, building your own second brain linked through files, or setting up a content pipeline from brief to publication are three projects of equal demand, only one is needed to pass the course.
The capstone imposes only one rule: deliver one of the three proposed projects all the way through, in the real environment, with a plan approved before execution. None of the three is valued more than the others, only one is needed to pass the course.
Three projects, one shared demand
The first project ships a small real feature in an existing or new project: a screen, an endpoint, a fix that genuinely changes something observable for someone who uses it. The second builds your own second brain, a folder of notes linked together through files, reread and enriched across sessions rather than restarted each time from a blank page. The third sets up a complete content pipeline, from the initial brief through to a publishable deliverable, reusing the steps seen in the content pipeline from the previous module.
All three carry the same underlying demand: an observable result, produced in a single session, checked on the spot rather than assumed finished.
Framing before executing
The chosen scope fits within a single working session, not a week of chained sessions: a project that is too broad rarely gets finished, a project scoped to one observable action gets finished and gets checked. The plan is approved before execution, even when no one else reads it straight away, because rereading your own plan before acting reveals forgotten steps faster than rereading only the finished result.
Périmètre annoncé : ajouter un filtre de recherche sur une liste existante
Geste central : le filtre renvoie les bons résultats sur un jeu de trois requêtes test
Preuve de réussite : capture de la liste filtrée, dans le navigateur réel, pas dans le seul diff de codeThe final check happens in the real environment: the application that runs, the file that opens, the content that actually publishes, never in the mere reading of the code written. A clean diff guarantees nothing about what happens once run on a real machine.
Anyone who wants to go further can move on to the multi agent adversarial review, a more demanding project that is not required to pass the course.
The three capstone projects, on the same criteria
| Capstone project | Expected scope | Central action | Expected proof of success |
|---|---|---|---|
| Real feature | A screen, an endpoint or a fix in an existing or new project. | Deliver an observable change for someone who actually uses the project. | The change observed running, not merely read in the source code. |
| Personal second brain | A folder of notes linked together through files, on a subject followed over time. | Link new notes to existing notes rather than writing into a void. | A note found and reused through its links, not merely a note written once. |
| Content pipeline | Content that goes from a written brief through to a publishable deliverable in its final format. | Take the content through every named step of the pipeline, without skipping any. | The final deliverable opened and checked in the format where it will actually be read. |
A person chooses the personal second brain project for the capstone. She scopes it to creating ten notes linked together, writes a plan in three steps, then shares it for review before starting to write the notes.
Write in one sentence what this situation establishes, and in one sentence what it does not establish.
What this establishes: This process establishes that she scoped her project to a precise number of notes and had her plan approved before execution, in line with the discipline expected for the three capstone projects.
What this does not establish: It does not establish that the ten notes, once written, will actually be usefully linked together in the real environment of the second brain.
The three most common miscalibrations
- Too broad This person has passed the course since her plan was approved before execution.
- Too narrow Nothing can be said until the ten notes have been reviewed one by one by a third party.
- Beside the point This situation shows that the second brain project suits an organised person better than the other two projects.
- The three projects, real feature, personal second brain, content pipeline, carry the same demand for validation, none counts as harder than the others.
- A scope framed to a single working session gets finished and gets checked, whereas a scope spread over several weeks rarely condenses into a finished deliverable.
- The project plan is approved before execution, which reveals forgotten steps earlier than rereading only the result.
- Proof of success is observed in the real environment where the content actually runs, never in the mere reading of the code diff produced.
Choose now, among the three projects described here, the one you are going to scope, then write in three sentences the exact scope you plan to deliver in a single working session.