Current work

Updated: 2026-09-08. The next product step is brainstorming actual assets and asset types. This baseline exists to make that discussion concrete; asset decisions may revise the program, minting factors, renderers and games.

Standalone promo sequence

Accepted by jpc as a good first pass. Keep it standalone and unlinked while content is refined. A watch concept beat is now included in the player as a visual test; integration into Moments/clipping or the homepage remains a future option.

/reel is a public direct-link page, excluded from site navigation and marked noindex. It has the accepted wordmark and one playback frame: animated text → first streamer clip → animated text → second streamer clip → animated text → standalone ownership clip → watch Look Shift (sealed → open/active → evolved) → closing text. No scroll-triggered sequencing. GSAP follows one HTML video clock, including pause/seek/buffering. A single Play with sound gesture starts the sequence; sound, pause, seek, restart and fullscreen controls remain available. Reduced motion removes depth effects.

Three original ElevenLabs clips are retained in public/media/reel/. A local FFmpeg playback assembly adds the silent watch interval after the third clip and extends the sequence to 31 seconds; CSS/GSAP render the intertitles and watch states in the page. The watch reference board and crops are in public/media/reel/watch/. This assembly uses no new generative-video credits. Watch value/ownership labels, like prices, chat and ownership in the AI concept clips, are illustrative; the first clip’s spoken-number error remains in the supplied audio. Main navigation, Moments, RPC reads and the Solana program are unchanged.

Initial expansion directions

• Keep the promo as one framed audit: text, a human clip or asset state, then a transition into the next beat. Do not turn it into a scrolling gallery.
• Treat the watch sequence as the first reusable asset module: a consistent object, a few visible states, a short Look Shift, and concise value/ownership context. Timing and placement remain adjustable after review.
• Pair future asset modules with clips that give them social context. The clip can establish the reaction or claim; the asset beat can show the decoded appearance, change, price movement or ownership moment.
• Candidate follow-up modules are the selected animated cartoon cat with progressive reveal followed by animation, an original unbranded car with its own flex-worthy transformation, and later a clearly grouped collection or real-estate concept. These are prompts for the next creative pass, not approved production assets.
• Keep generated art, CSS/GSAP transitions and future video inserts interchangeable. Each module should work as a standalone moment on /reel before we decide whether it belongs in Moments or the homepage.
• Continue marking concept values, owners, chat and event language as illustrative until those states are connected to the relevant Solana records and approved product rules.

Implemented baseline

Program v0.3.0 was compiled here and upgraded on verified Solana Devnet. It enforces the W VIBE issuer, derives evolution-dominant multi-factor rarity and initial test prices, and binds acquisition/game inputs at issuance. V1/V2 records retain their meanings. See Minting for exact authored inputs, formula and scope.

The active 100-record batch is reissued under V3 with predecessor bindings, three temporary image types, five cadences, eight Stashes, 61 public records and 39 hidden records. Thirty-five Dubs are featured across three pages. The sample distribution is two minute, eight hourly, twenty daily, thirty monthly and forty yearly records. These are observed batch counts, not global supply guarantees.

The Chase has three type-selected games: Artifact/Rebuild, Entity/Resonate and Scene/Pathfinder. Reward terms distinguish designated free claims, reveal only, disabled offers and excluded inventory. The program enforces claim eligibility; the application validates game progress. Five designated free rewards remain unclaimed in this starting batch. Purchases and discounts remain disabled.

A real wallet-signed claim was confirmed through the published Site and its original owner/finder and Moment verified. Its original record remains intact; the refreshed counterpart retains that owner. Signed simulations separately verify V3 free-claim success and offer/reveal refusal. See Status for the exact evidence and verification limits.

Operational rules

jpc confirmed the deployed RPC retry/refresh fix works. The intermittent loading incident affected the homepage, Moves and Moments, which share the repaired read path. Production evidence established an RPC timeout; caching was not established as its cause. See Status and Development.

Shared tooling was restored on 2026-09-07 outside /workspace: Rust/Cargo in /root/.cargo and /root/.rustup, Solana under /usr/local/lib/solana. /root/.bashrc and /root/.profile load the shared PATH. Stable Rust 1.98.1 is the general default; W VIBE explicitly uses SBF platform tools v1.57. All eight Rust tests passed and the rebuilt executable exactly matched the saved release. This installation did not deploy anything. Check executable paths/versions again when resuming; installations may disappear independently of project files. See Development. Missing local tooling does not imply a Devnet reset.

Do not use the internal preview browser. This is the user's explicit workflow preference. Publish coherent authorized checkpoints and test the published Site APIs over HTTP against real Devnet. Do not retry cloud-browser navigation to published Sites URLs.

No D1, inventory TTL or persistent snapshot cache is used. Reads report chain time, slot range and generation, and run independently within a 35-second server budget. The browser allows 40 seconds. Each read-only RPC has a 5-second attempt deadline and one transient retry. The browser retries its first failed read after 1.5 seconds, then uses the normal 30-second cadence; valid same-query data stays visible with its timestamp on refresh failure. Restart read supersedes pending work, and background/BFCache restoration is handled. Hidden details/events remain excluded server-side. Keys remain private. Recovery is administrative, with explicit missing-state checks and checkpointed replay; a timeout is not evidence of a Devnet reset.

Current workshop position

Watches have completed their first pass. Working direction: no Discovery for watches; original unbranded designs, likely premium acquisition, with mechanism, interactivity, evolution and case presentation unresolved. One-second evolution is only an option. See Asset workshop.

Cats have completed their first pass: animated cartoon cats, progressively revealed, then animation begins. Realistic cat pictures are excluded. Style, reveal timing/stages, behavior, sound, traits/rarity and implementation need a later pass. Cat Discovery and acquisition remain undecided.

Next: original, unbranded cars. Continue short first-pass conversations across real estate and the remaining candidate categories, then revisit and refine the set before implementing. No connected virtual world, geographic mirror or general creatures category is planned. Premium collections group disclosed rare assets and need not be geographic. Ownable parent assets and sub-assets require a separate model; current Stashes are flat groupings.

The current puzzles are technical tests and are not the accepted discovery design. Do not expand them into the proposed product or implement watch discovery. Record the discussions; wait for asset behavior and acquisition choices before changing minting, program logic, rendering or games.