Skip to content

docs: add product roadmap (Harbor storage + Talus on-chain direction)#4

Open
harrymove-ctrl wants to merge 2 commits into
mainfrom
docs/product-roadmap
Open

docs: add product roadmap (Harbor storage + Talus on-chain direction)#4
harrymove-ctrl wants to merge 2 commits into
mainfrom
docs/product-roadmap

Conversation

@harrymove-ctrl

Copy link
Copy Markdown
Owner

What

Adds docs/ROADMAP.md — a single, grounded product roadmap covering the next 0-12 months across six areas: Walrus/Harbor storage, smart-contracts/on-chain architecture, crawling/extraction, UI/UX, onboarding, and product completeness. The roadmap is aligned to the Talus/acc grant goal of publishing contextmem.extract/.recall/.remember as Nexus DAG tools with on-chain attribution receipts.

Why

ContextMEM has no custom on-chain layer today: storage is Tatum-gateway (plaintext, mainnet, Node-only, never in the Worker), "private" namespaces are a D1 token check rather than encryption, and there is zero Move code, zero attribution receipts, and zero Talus/Nexus wiring. The roadmap sets a clear direction and sequences the work to ship the grant demo and harden for public launch.

Structure

  • Executive summary and an honest current-architecture snapshot (no Move contract; Tatum-based, plaintext Walrus; MemWal-pointer memory; app-layer namespaces).
  • A cross-cutting Harbor + on-chain decision section up top with a clear recommendation.
  • The six areas, each with current state → gaps → a prioritized plan table (Task | Priority | Effort | Notes) → Harbor/on-chain notes → open questions.
  • A phased timeline mapping P0/P1 work to Talus milestones (M1 0-3mo / M2 3-6mo / M3 6-12mo).
  • A consolidated Decisions needed from the team list.

Key recommendations

  • Coexist, don't replace: keep Tatum for public/mainnet; add Harbor (Sui testnet, SEAL-encrypted) for private namespaces + the on-chain demo, until Harbor reaches mainnet.
  • Hybrid on-chain (Option C): reuse Walrus + Harbor/SEAL + MemWal, and author one minimal Move package (registry + receipt) for typed, composable attribution receipts — the one thing Harbor's Bucket/Seal objects don't provide.
  • Harbor lands before any Move authoring — its Bucket + Seal policy objects (created via Enoki-sponsored, no-SUI-balance txs) are the first real on-chain layer.
  • Spike @mysten/seal in the Worker early — the key runtime unknown that decides where encryption runs.
  • Build the usage/event ledger first — the shared spine for billing, observability, and on-chain attribution.

Docs-only change; no code or behavior is modified.

Plans the next 0-12 months across six areas — Walrus-memory+Harbor,
on-chain/smart-contract architecture, crawling/extraction quality,
UI/UX, onboarding, and cross-cutting feature gaps — grounded in the
current codebase and aligned to the Talus/acc milestones.
This was referenced Jun 16, 2026
…Namespace, web split, CI scope)

Fact-checked file:line/symbol/architecture claims against the codebase:
- on-chain reads: getObject only in proof.ts/resources.ts; history.ts raw JSON-RPC, resolve.ts SuiNS lookupName, bcs.ts BCS schemas
- CloudflareNamespaceStore.authorizeNamespace at worker.ts:502 (was .authorize :503)
- apps/web is split into components/ + lib/ (not only the constellation)
- CI runs typecheck+test+build (+ gated deploy.yml), not typecheck+test only
@cloudflare-workers-and-pages

cloudflare-workers-and-pages Bot commented Jun 19, 2026

Copy link
Copy Markdown

Deploying contextmem with  Cloudflare Pages  Cloudflare Pages

Latest commit: 6ed3e78
Status: ✅  Deploy successful!
Preview URL: https://c09b84bd.contextmem.pages.dev
Branch Preview URL: https://docs-product-roadmap.contextmem.pages.dev

View logs

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants