docs(ai): add lessons migrated from claude memory

[caseyisonit]

Based on #6788 (layer 4), then #6787, #6786, and #6785. This PR's diff is relative to layer 4.

Description

Adds three lessons from my Claude memory to .ai/memory/agnostic-lessons.md, so every agent that reads .ai/ gets them, not only Claude.

• Styling API (from feedback-swc-styling-api-prefix): gen2 components expose --swc-* custom properties as their public styling API, for example --swc-badge-height in Badge.ts. --mod-* and --spectrum-* are 1st-gen and Spectrum CSS conventions and belong in gen2 docs only in migration-guide comparisons. The lesson notes that Stylelint's custom-property-pattern accepts any lowercase kebab-case name, so the prefix isn't linted.
• Generated files (from feedback-update-stale-generated-diffs): when yarn build rewrites generated output, such as gen2/packages/swc/stylesheets/global/global-*.css from gen2/packages/tools/vite-global-elements-css, commit the regenerated file even if the diff looks unrelated. Revert it only if it contradicts its source.
• Documentation (from feedback-no-hardwrap-markdown-docs): don't hard-wrap prose in .md and .mdx. .prettierrc.yaml leaves proseWrap at its default, preserve, so Prettier never reflows hand-wrapped lines.

The --swc-* lesson lives in agnostic-lessons rather than css-styling-lessons because it governs MDX docs, @cssprop JSDoc in .ts files, and examples. css-styling-lessons loads only for **/*.css, so it would never be in context while editing those files.

The frontmatter description now names the two new categories. yarn ai:sync regenerated the Copilot instruction, the Cursor rule, and the .ai/README.md catalog entry. No other files were edited by hand.

Not migrated:

• Comment-brevity standard: needs team agreement before it becomes repository guidance.
• CEM tooling memory: stale, because @pwrs/cem isn't used in gen2.
• Icon RFC memory: stale, because icon-rfc.md no longer exists.
• Personal goals and feedback: personal context, not repository guidance.
• Docs-structure and migration-tooling plans: RFC material, not lessons.

Motivation and context

These lessons came from review corrections, but they lived only in one person's Claude memory, so Copilot, Cursor, and other contributors' agents didn't get them. Moving them into .ai/memory/ makes them path-scoped instructions that every supported tool loads.

Related issue(s)

• Stacked on docs(ai): rewrite .ai readme around a generated catalog #6788, refactor(ai): normalize skills and retire .ai/config.json #6787, feat(ai): split stories rules and load memory as instructions #6786, and feat(ai): generate copilot and cursor instructions from .ai rules #6785. This is part of the AI configuration migration plan.

Screenshots (if appropriate)

Not applicable: no visual changes.

---

Author's checklist

• I have read the CONTRIBUTING and PULL_REQUESTS documents.
• I have reviewed at the Accessibility Practices for this feature, see: Aria Practices. Not applicable: no UI changes.
• I have added automated tests to cover my changes. yarn lint:ai checks the lessons frontmatter and generated-file drift.
• I have included a well-written changeset if my change needs to be published. It doesn't: no published package changes.
• I have included updated documentation if my change required it.

---

Reviewer's checklist

• Includes a Github Issue with appropriate flag or Jira ticket number without a link
• Includes thoughtfully written changeset if changes suggested include patch, minor, or major features
• Automated tests cover all use cases and follow best practices for writing
• Validated on all supported browsers
• All VRTs are approved before the author can update Golden Hash

Manual review test cases

• Generated files are in sync

  1. Run yarn ai:sync --check, then yarn lint:ai.
  2. Expect both to pass.

• Copilot loads the lessons

  1. Run copilot instruction list.
  2. Expect memory-agnostic-lessons.instructions.md from .github/instructions/.

• Each lesson is in the generated instruction

  1. Search .github/instructions/memory-agnostic-lessons.instructions.md for ## Styling API, ## Generated files, and Don't hard-wrap prose.
  2. Expect one match for each.

Verified by the author: yarn ai:sync --check, yarn lint:ai, Prettier, and copilot instruction list pass, and each lesson appears once in the generated instruction.

Device review

• Did it pass in Desktop?
• Did it pass in (emulated) Mobile?
• Did it pass in (emulated) iPad?

Accessibility testing checklist

This PR changes only agent instruction files. It adds and changes no components, stories, or rendered UI.

• Keyboard: not applicable. There are no focusable or interactive parts to test.
• Screen reader: not applicable. There is no rendered content to announce.

[changeset-bot]

⚠️ No Changeset found

Latest commit: 81cc95f

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[github-actions]

📚 Branch Preview Links

• Storybook (Gen2, dev mode)
• Storybook (Gen2, prod mode)
• Documentation Site (Gen1)
• Storybook (Gen1)

🔍 Gen1 Visual Regression Test Results

When a visual regression test fails (or has previously failed while working on this branch), its results can be found in the following URLs:

• Spectrum | Light | Medium | LTR
• Spectrum | Dark | Large | RTL
• Express | Light | Medium | LTR
• Express | Dark | Large | RTL
• Spectrum-two | Light | Medium | LTR
• Spectrum-two | Dark | Large | RTL
• High Contrast Mode | Medium | LTR

Deployed to Azure Blob Storage: pr-6806

If the changes are expected, update the current_golden_images_cache hash in the circleci config to accept the new images. Instructions are included in that file.
If the changes are unexpected, you can investigate the cause of the differences and update the code accordingly.

[cdransf]

Looks good! ✨