# BLOCKS.md **PointCast v2.1 — the Blocks pivot, live edition** Author: Mike Hoydich × Claude Revision: **v2.1** (2026-04-17) — live on `main`, serving pointcast.xyz. Previous revisions: - **v2.1** (2026-04-17 evening) — adds 9th channel BTL (Battler), `media.thumbnail` + `media.ipfsFallback` schema fields, agent-layer surfaces (/manifesto, /glossary, /agents.json, /llms-full.txt), home-page majors strip (inline /drum + /cast), stronger per-type variants + scroll rhythm via `:nth-child`, BreadcrumbList + ItemList JSON-LD coverage. - **v2.0** (2026-04-17) — mainnet origination of Visit Nouns FA2 at KT1LP1oTBuudRubAYQDErH7i7mSwazVdohxh, 10 starter mints, TZIP-21 metadata endpoint live, /collection page. - **v2.0-rebuild** (earlier April 2026) — initial Blocks pivot off v1. 8 channels, 8 types, monotonic IDs, Astro content collections. - **v1** (2025 → April 2026) — markdown-doc era; retired at commit 7fea01c, preserved for rollback. Original dispatches, drum room, Noun Bells instrument, /collect inventory. --- ## Why this document exists pointcast.xyz v1 works but its visual language reads as "well-considered content site" — markdown-doc rhythm, generous whitespace, understated type. Not distinctive enough. v2 rebuilds around a single primitive (the Block) with a grammar dense enough to own a visual niche, agent-native structure, and Tezos-native mint/faucet mechanics. Do not preserve v1's layout. Do preserve: Nouns IP usage via noun.pics, the visit log concept, the dispatch archive, the content itself. --- ## The Primitive — Block Every piece of content on PointCast is a Block. One primitive, stable grammar, varied content. ### Block schema ```ts type Block = { id: string // "0205" — monotonically increasing, zero-padded to 4 digits channel: Channel // see Channels below type: BlockType // see Types below title: string // short, sentence case body?: string // markdown, optional timestamp: ISO8601 // when published size: "1x1" | "2x1" | "1x2" | "2x2" | "3x2" // desktop grid footprint; default "1x1" noun?: number // Nouns seed ID for the block's icon edition?: { supply: number | "open" minted: number price: { tez?: number; usd?: number } | "free" chain: "tezos" // Tezos-only in v2; Ethereum optional later contract: string // KT1... address tokenId: number marketplace?: "objkt" | "fxhash" | "teia" } media?: { kind: "image" | "audio" | "video" | "embed"; src: string } external?: { label: string; url: string } meta?: Record // free-form agent-readable tags } ``` Every field except `id`, `channel`, `type`, `title`, and `timestamp` is optional. Blocks are JSON files in `/content/blocks/{id}.json` — no database, git is the database. ### Channels Ten channels. Each has a code, a color, and a purpose. Do not add an eleventh without Mike's decision. | Code | Name | Color (hex / ramp) | Purpose | |-------|-----------------|--------------------------|--------------------------------------------| | FD | Front Door | `#185FA5` / blue-600 | AI, interfaces, agent-era thinking | | CRT | Court | `#3B6D11` / green-600 | Pickleball — matches, paddles, drills | | SPN | Spinning | `#993C1D` / coral-800 | Music, playlists, listening notes | | GF | Good Feels | `#993556` / pink-800 | Cannabis/hemp, product drops, brand ops | | GDN | Garden | `#0F6E56` / teal-600 | Balcony, birds, wildlife, quiet noticing | | ESC | El Segundo | `#534AB7` / purple-600 | ESCU fiction, local, community | | FCT | Faucet | `#BA7517` / amber-600 | Free daily claims, giveaways | | VST | Visit | `#5F5E5A` / gray-600 | Human/agent visit log entries | | BTL | Battler | `#8A2432` / oxblood-600 | Nouns Battler — deterministic Nouns duels | | BDY | Birthday | `#D86440` / coral-600 | Birthdays celebrated on PointCast (indexed at /cake) | Channel code appears in monospace as `CH.{CODE} · {ID}` at the top of every block. ### Block types Types govern internal treatment. Not about-ness (that's channel), but form. | Type | What it is | Footer shows | |----------|--------------------------------------|----------------------------| | READ | Long-form text, essay, dispatch | Reading time · ED count | | LISTEN | Spotify/SoundCloud/audio embed | External link | | WATCH | Video embed | External link · duration | | MINT | Paid edition, collectible on Tezos | Price in tez · supply · button | | FAUCET | Free claim, daily reset on Tezos | Claimed · cap · button | | NOTE | Short observation, tweet-sized | Location tag | | VISIT | Someone stopped by and signed log | Agent vendor or geo | | LINK | External link with context | Destination domain | | TALK | Voice Dispatch — 10-60 sec audio (RFC 0001) | Duration | | BIRTHDAY | Open-edition card, one person per year, free FA2 | Recipient · Noun · claimed | Ten channels × ten types = 100 combinations. Most will never appear. That's fine. --- ## URL structure Agent-legible, human-legible, permanent. ``` / # home — block grid, most recent first /b/{id} # single block page (e.g. /b/0205) /b/{id}.json # machine-readable block /c/{channel} # all blocks in a channel (e.g. /c/front-door) /c/{channel}.rss # RSS feed per channel /c/{channel}.json # JSON feed per channel /blocks.json # full archive, paginated /sitemap-blocks.xml # every block, for crawlers /for-agents # manifest page — who this site is, how to read it /visits # the log, as before /visits.json # structured visit log ``` Block IDs are immutable. Never reuse, never renumber. If a block is retired, it 404s — the ID does not get handed to something else. --- ## Visual language ### What to pull from - **Bloomberg Terminal** — channel color as primary navigation, monospace metadata, density without clutter, hard-edged panels - **Nouns.wtf** — numbered-forever discipline, Noun as first-class visual identity per block - **Warpcast / Farcaster** — card primitive with embeds and edition metadata, mint buttons inline with content - **objkt / fxhash** — edition counts visible as primary metadata, not hidden in a modal - **Teletext / Prestel / CEEFAX** — numbered pages as an addressing system, tight type, agent-native by accident - **Vintage trading cards** — the block feels like an artifact you could hold ### What to avoid - Generous whitespace for its own sake - Understated greys and muted palette - Serif blog aesthetic (Medium, Substack default) - Notion-esque doc rhythm - Anything that reads as "minimal SaaS" - Navigation sidebars (the channel colors ARE the navigation) ### Typography - **Monospace for metadata** — channel codes, IDs, timestamps, edition counts. Use `JetBrains Mono`, `IBM Plex Mono`, or `Berkeley Mono` via self-hosted fonts. Never system mono — it reads as code, not as broadcast signature. - **Sans for titles and body** — `Inter` as default. Consider `Neue Haas Grotesk Display` for dispatch titles if the licensing budget allows later (deferred). - **Letterspace all-caps monospace metadata at 0.08–0.12em**. That wide-tracked small-caps look is half the signature. - **Two weights only**: 400 and 500. No 600/700 — too heavy for this density. - **Sentence case** for titles. Caps-lock only for metadata. ### Color application - Block background: white or very light tint of the channel color - Block border: 1.5px solid in the channel's 600-stop - Left accent bar: optional 6px solid channel color for reading-type blocks - Channel code text: channel's 800-stop for contrast - Never mix more than two ramps in one block. Visual noise kills density. ### Spacing and structure - Grid gap: 8px (tight — blocks should feel like a wall of signal, not a gallery) - Block internal padding: 14–16px - Border radius: 0 or 2px maximum. Hard corners are part of the language. No rounded pills anywhere except inline badges. --- ## Layout behavior ### Desktop (≥1024px) - CSS Grid, `grid-template-columns: repeat(auto-fit, minmax(220px, 1fr))`, `grid-auto-flow: dense` - Blocks claim their footprint via `size`: `1x1` is the default; `2x1`, `1x2`, `2x2`, `3x2` as needed - Header: wordmark + block count + current time, one line - Footer: channel legend + endpoint list - No sidebar. No top nav. The grid is the navigation. ### Mobile (<768px) - Single column, full-bleed - Every block gets equal horizontal space regardless of `size` - Gap becomes 12px vertical - Swipe-scrolling, not pagination - Channel filter as a sticky horizontal chip bar at top — tap to filter ### Tablet (768–1023px) - Two columns. `size` downgrades: `2x1` becomes `1x1`, `2x2` becomes `2x1`. ### Agent mode - When `User-Agent` matches known agent patterns (GPTBot, Claude-Web, Perplexity, Atlas, etc.), serve a stripped HTML version: no CSS, no JS, pure semantic markup with rich JSON-LD blobs - Also serve `/for-agents` as a human-readable manifest explaining the site structure, endpoint list, and preferred citation format --- ## Faucet and mint — Tezos-first implementation **Why Tezos first**: the existing objkt / fxhash / Teia culture is the audience PointCast is built for — artist-run publications, cc0 gen-art, thoughtful curated drops. Ethereum NFT culture in 2026 is speculation-heavy; Tezos is where taste-meets-machine actually lives. Mike is already present on objkt as `mhoydich`, so the audience doesn't need to be rebuilt. Mint costs on Tezos are a fraction of a tez (pennies), making the daily faucet economically viable in a way Ethereum L1 never was. ### Technical stack - **Token standard**: FA2 (Tezos-native multi-token standard) - **Contract language**: SmartPy (preferred) or LIGO — whichever the contract-in-progress uses. Stay consistent. - **Wallet protocol**: Beacon (Tezos standard) - **Supported wallets**: Temple, Kukai, Umami at minimum. Beacon handles all three. - **RPC provider**: Tezos public nodes (SmartPy, ECAD, TzKT) or dedicated provider if rate limits bite - **Indexer**: TzKT API for reading mint counts, holder lists, marketplace listings - **Marketplace surfaces**: objkt primary, fxhash for gen-art drops, Teia optional ### Faucet (ship first) - **Mechanic**: one tokenId per day, resets at 00:00 PDT, supply cap per day (start at 50 or 100) - **Claim cost**: gas only (~0.01 tez, effectively free) - **Limit**: one claim per wallet per day (enforced in contract) - **Metadata**: each day's noun is the token artwork, stored on IPFS via a pinning service (Pinata or NFT.Storage) - **Block surface**: `claim()` button on Faucet-type block, reads live `minted/supply` from TzKT, disables if user already claimed today or supply exhausted - **No wallet required to browse** — only to claim ### Mint (ship second) - **Same contract structure**, paid editions - **Dispatches** can mint as collectible editions (optional, not every dispatch needs this — Mike's call per dispatch) - **Good Feels drops** are phygital — mint returns a redemption code (off-chain, served by Good Feels backend) for in-store or DTC pickup - **Marketplace mirror**: minted tokens automatically appear on objkt and fxhash (FA2 standard, indexed natively). Block footer surfaces objkt link for secondary trading. - **Pricing**: denominated in tez, with live USD equivalent shown via TzKT price feed ### Ethereum path (deferred, optional) Not in v2. Ship Tezos clean, prove the mechanic works, watch the reception. If later there's clear demand from a specific audience segment that lives on Base or Zora, mirror specific drops cross-chain. Do not split the primary flow. Tezos is the home chain. --- ## Implementation phases ### Phase 1 — Visual rebuild (weekend 1) - New branch `blocks-rebuild` - Implement Block schema and rendering - Migrate existing v1 content into `/content/blocks/*.json` (preserve IDs where possible — assign sequential IDs otherwise) - Build the home grid and single-block pages - Ship to a preview URL, not production, until ready ### Phase 2 — Agent layer (half-day) - `/for-agents` manifest - JSON and RSS feeds per channel - JSON-LD on every block - `sitemap-blocks.xml` - Agent-detection middleware for stripped-HTML mode ### Phase 3 — Tezos faucet (timing depends on in-progress contract work) - Finish / deploy FA2 contract on Tezos mainnet (currently in progress — coordinate with whichever agent is leading this) - Beacon wallet connect on the frontend - Daily claim mechanic with per-wallet rate limit - Faucet block type wired to contract + TzKT indexer - IPFS pinning pipeline for daily noun metadata ### Phase 4 — Paid mints - Extend FA2 contract (or deploy companion) for paid editions - Good Feels phygital redemption flow (redemption code lives in Good Feels backend; PointCast block displays it after mint verification) - objkt listing automation ### Phase 5 — Cutover - Migrate pointcast.xyz to v2 - v1 archived at a subdomain if sentimental value warrants it --- ## Out of scope - Comments, likes, social features of any kind - User accounts beyond wallet connect - Templating for other people to use PointCast - Email newsletter (RSS + JSON feeds are the newsletter) - Mobile app - Analytics beyond simple visit counts - Ethereum / Base minting in v2 (deferred) --- ## Open questions for Mike 1. **Contract status** — how close is the current Tezos contract to deployable? SmartPy or LIGO? Who's leading the contract work currently (Claude Code, or parallel Manus/Codex effort)? 2. **Nouns IP** — keep pulling from noun.pics for block imagery, or commission a custom PointCast-specific noun set? 3. **Typography budget** — Inter free works for v1; defer Söhne / Neue Haas to v2.1? 4. **Good Feels drops** — contract lives on PointCast's FA2 or a separate Good Feels collection? 5. **Daily faucet noun selection** — random pull from noun.pics, curated rotation, or custom daily drops? 6. **Marketplace strategy** — objkt only, or also fxhash and Teia from day one? --- *The PRD from January 5 described the thesis. This document describes the execution. Every design decision must serve the Block primitive. When in doubt, make it denser, more numbered, more agent-legible, and more visually distinct from a default Notion page. Tezos is the home chain.*