Keep each artifact role visible
The synthetic memory set gives each source, first attempt, finding, correction, and final record a role that remains clear later.
Build names from stable parts
Use a short sequence, a role, and an optional subject. For example:
- 01-source-notes-workshop.md
- 02-first-brief-workshop.md
- 03-independent-finding-workshop.md
- 04-corrected-brief-workshop.md
- 05-final-summary-workshop.md
The sequence shows reading order. The role explains why the file exists. The subject separates this job from another one without packing every detail into the name.
Keep file names readable and portable
CommonMark defines a plain-text format that can remain readable across many tools. W3C PROV-O describes relationships among entities, activities, and agents, which helps explain why role distinctions matter. RFC 8493 describes a way to package files with an inventory when the set must travel together.
Names should support those relationships, not carry the entire history. Put dates, sources, and decision details inside the artifact when they would make the filename unwieldy.
Know where naming stops
Role-based names fit recurring work with several related artifacts. They add little to a disposable note that will never support another decision.
Naming does not decide which version is current. It also does not verify that a copied folder is complete. Those are separate choices with separate checks.
Continue through the memory set
Start with Keep AI Work in Files You Control for the owned-record decision. Next, Mark the Current AI Work Artifact separates the chosen brief from earlier attempts. Verify an AI Workspace Copy covers completeness after files move.