Trustless in-tab full node: run skycoin/fibercoin daemon in the wasm visor (storage-backend + signed-snapshot fast-sync)

[0pcom]

Vision

Run a full skycoin/fibercoin node inside the wasm visor (browser tab) — a trustless in-tab wallet that verifies the chain locally instead of querying a remote full node it doesn't control. Now that skywire + dmsg run in the browser (demonstrated by the skycoin-web thin-client integration over dmsg), the networking constraint that made this unthinkable has evaporated: the tab already talks peer-to-peer over the mesh. What's left is storage + sync — an implementation problem, not a physics one. Likely a first-of-its-kind (browser full node for a UTXO chain).

Why it's worth it (the value, not just the flex)

• Privacy: querying a remote full node for balances/history leaks which addresses you care about to a node you don't control. A local node answers those queries against your own verified copy — nothing leaks. (We can still offer the remote path too — connect over dmsg to a full node you control — but local is the strongest privacy posture.)
• Trustlessness: verify the chain yourself rather than trusting a remote node's answers.
• Abstraction fit: the daemon is "external" today only because of its storage engine, not because of what it is. Swap the backend and it becomes another internal-app RunFunc with a wasm body — the exact pattern skycoin-web now uses, one layer deeper (the seam moves from "serve the SPA" to "what's my KV backend").

The real constraints (and why they're tractable)

• bbolt is the wall. It needs mmap + flock + POSIX file syscalls; js/wasm has no filesystem and wasip1 mmap/lock support is flaky. Stock storage is DOA in a tab. → Swap the backend behind skycoin's DB interface to an OPFS- or IndexedDB-backed KV store. (SQLite already ships a wasm build on an OPFS VFS — persistent DB-in-a-tab is proven, shipping tech.)
• wasm32 4 GB address space → can't hold the ~700 MB chain all in RAM; need a paged, on-disk-shaped store (→ OPFS), not in-memory.
• Sync/verify cost. skycoin's single-block-publisher consensus has no PoW to replay — verification is signature-checking + tx/UTXO rules, far cheaper than a PoW browser node. This also enables fast-sync from a signed UTXO snapshot (trust the publisher's signature from a checkpoint forward), collapsing both sync time and footprint.
• Fibercoins (small/fresh chains) are feasible sooner — a tiny chain could run a full node in a tab today, even in-memory. skycoin mainnet is the harder case but not a wall.

Rough path (research-grade, not near-term)

1. Abstract skycoin's block/UTXO DB behind a KV interface (if not already).
2. OPFS/IndexedDB-backed KV backend for the wasm build (mirror the SQLite-wasm-on-OPFS approach).
3. Signed-snapshot fast-sync (checkpoint UTXO set + verify forward) — leverage the block-publisher signatures.
4. Register skycoin-daemon as an internal app with a wasm RunFunc (same launcher path as skycoin-web / skychat), backed by the wasm KV store; blocks fetched over dmsg.
5. Local wallet queries hit the in-tab node; keep the dmsg-remote-node option for users who prefer it.

Status

Deliberate detour, not near-term. Keep the daemon external for now (thin-client-over-dmsg is the default). Filing so the design + rationale aren't lost — the day someone wants a trustless in-tab node, skycoin is the chain most likely to make it actually work, and the internal-app abstraction means the daemon can graduate later without breaking the model.

Related: skycoin-web internal-app integration + wasm Apps-tab convergence (the abstraction this builds on).