feat(sdk): stage 5 — export adapter factories from package root

[vanceingalls]

Stage 5 — public adapter factory exports

Ships the public entry points for the SDK's adapter layer. The concrete adapters already existed internally (memory.ts, headless.ts, fs.ts) and pass the T13 contract suite, but index.ts only re-exported the type interfaces — so consumers had to reach into deep src/adapters/* paths to instantiate one. This wires them to the package root.

What's exported

Factory	Adapter	Use case
createHeadlessAdapter()	PreviewAdapter (no-op)	Agents, CI, SSR — no browser, no DOM
createMemoryAdapter()	PersistAdapter (in-memory)	Tests, demos, headless open with no storage backend
createFsAdapter(opts)	PersistAdapter (node fs)	Local dev — versioned history on disk

FsAdapterOptions is exported alongside createFsAdapter.

Why this completes the headless story

Agents and CI can now open a composition with zero browser and zero disk:

import { openComposition, createHeadlessAdapter, createMemoryAdapter } from "@hyperframes/sdk";

const session = await openComposition(html, {
  preview: createHeadlessAdapter(),
  persist: createMemoryAdapter(),
});

Scope

• S3 / production HTTP adapters are post-Pacific — intentionally not included.
• No new adapter implementations; this is purely the public export surface.

Verification

• index.ts only, +6 lines
• oxlint clean, oxfmt clean
• sdk typecheck clean
• 156/156 sdk tests pass

🤖 Generated with Claude Code

[vanceingalls]

• feat(sdk): stage 6 — sub-composition scoped ids (F9) #1434 : 2 dependent PRs (#1429 , #1441 )
• feat(sdk): stage 5 — export adapter factories from package root #1432 👈 (View in Graphite)
• feat(sdk): stage 4 — canUndo/canRedo, removeElement GSAP cascade, override-set cleanup #1431
• feat(sdk): can() returns CanResult; T4 dispatch-boundary tests #1426
• fix(sdk): fs adapter flush() tracks in-flight writes; add to T13 contract suite #1425
• feat(sdk): file-backed fs adapter + setTiming GSAP-script sync; add sdk-playground #1423
• refactor(core): swap studio-api read path from recast to acorn parser (T6e) #1392
• feat(core): parse-parity suite for acorn parser (T6d) #1370 : 1 other dependent PR (#1379 )
• feat(core): acorn GSAP write path — magic-string offset-splice (T6c) #1369
• feat(core): acorn GSAP read path with T6b differential corpus tests #1368
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[james-russo-rames-d-jusso]

Verified at HEAD 7da2e36.

Trivial PR (+6 lines, single file). Adds createMemoryAdapter / createHeadlessAdapter / createFsAdapter + FsAdapterOptions to the package root entry. Diff is mechanical and the body explains the rationale clearly (the concrete adapters existed but only deep-paths could instantiate them).

Concern — stack-coherence flag (heads-up, not blocking this PR)

I cross-read sibling PRs in the same Graphite stack to verify import expectations. Two relevant observations:

• #1441 (http persist adapter) adds createHttpAdapter to the same root index alongside this PR's exports — wire-coherent, no conflict.
• #1443 (wire SDK session into Studio) removes createFsAdapter and FsAdapterOptions from the root index with the comment // Concrete adapter factories (browser-safe — Node-only fs adapter: @hyperframes/sdk/adapters/fs). and switches consumers to import from the @hyperframes/sdk/adapters/fs deep subpath instead.

So this PR's createFsAdapter root-level export is transient — it'll be reverted 5 PRs downstream in the same stack. The deep subpath ./adapters/fs already exists in packages/sdk/package.json on main, so consumers have a stable path either way.

That's not a blocker for #1432 — the rationale ("agents and CI can now open with zero browser") is real for the in-between PRs (#1434, #1429), and the headless + memory exports survive. But worth aligning intent: either:

• (a) skip createFsAdapter from the root in this PR to match the eventual state in #1443, or
• (b) leave as-is and accept the churn (which is what the current stack does).

Either is fine. Author's call.

No other findings. Mechanical export PR, oxlint/oxfmt/typecheck clean per the body, 156/156 tests pass.

Determinism: re-export only, no runtime code. N/A.

Verdict: approve.

Review by Rames D Jusso

[jrusso1020]

Approved.

[miguel-heygen]

Tiny re-export PR — no logic changes. CI blocked by root oxfmt failure in #1423.

The packages/sdk/src/index.ts comment clarifying that createFsAdapter is Node-only and must be imported from @hyperframes/sdk/adapters/fs (not the browser-safe root) is exactly right. Good that createHttpAdapter now exports from the root since it's browser-safe.

Verdict: APPROVE (pending #1423 oxfmt fix)
Reasoning: Export cleanup is correct and the Node/browser split is now explicit.

— magi

The base branch was changed.