diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 6da690d1..2b72bda3 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -206,8 +206,10 @@ jobs: with: path: .next/cache key: next-${{ runner.os }}-${{ hashFiles('bun.lock', 'next.config.ts') }}-${{ steps.next_cache.outputs.day }} + - name: Install the lockfile-pinned Chromium and Linux dependencies + run: node node_modules/playwright-core/cli.js install --with-deps chromium # Both browser suites start their own server on a free port against the - # same build and use the runner's preinstalled Chrome. They run one after + # same build and use the lockfile-pinned Chromium. They run one after # the other: the cross-tab contracts are timing-sensitive and a second # server plus browser contexts on the same runner made them flake. - name: Browser contracts against the production build diff --git a/app/messaging.ts b/app/messaging.ts new file mode 100644 index 00000000..391ee6fd --- /dev/null +++ b/app/messaging.ts @@ -0,0 +1,23 @@ +import snapshot from "../portfolio-messaging.generated.json"; + +if (snapshot.contract !== "hraness.product-messaging/v1" || snapshot.productId !== "aicharts") { + throw new Error("The website requires the canonical aicharts messaging snapshot."); +} + +export const productMessaging = snapshot.messaging; +export const productName = productMessaging.names.name; +if (!snapshot.canonicalUrl.startsWith("https://")) throw new Error("The product origin must use HTTPS."); +export const productCanonicalUrl = snapshot.canonicalUrl as `https://${string}`; + +/** Keep the site's selected siblings while deriving their copy and destinations. */ +export function relatedProduct(id: string) { + const project = snapshot.projects.find((candidate) => candidate.id === id); + if (project === undefined) throw new Error(`Unknown portfolio product ${id}.`); + return { href: project.canonicalUrl, name: project.name, role: project.messaging.short }; +} + +export function registeredRelationship(a: string, b: string): string | null { + return snapshot.relations.find((relation) => + (relation.source === a && relation.target === b) || + (relation.source === b && relation.target === a))?.detail ?? null; +} diff --git a/app/page.tsx b/app/page.tsx index ca3adfd0..ddc1249b 100644 --- a/app/page.tsx +++ b/app/page.tsx @@ -26,6 +26,8 @@ import { site, } from "./site"; +import { productMessaging, relatedProduct, registeredRelationship } from "./messaging"; + import "@/styles/chart-home.css"; export const metadata = createPublicSiteMetadata(searchSite, { canonicalPath: "/" }); @@ -63,7 +65,7 @@ export default function Home() {
-

Subscription vs API vs GPUs

+

{productMessaging.headings["home-calculator"]}

One maxed ChatGPT Pro seat implies a monthly token volume. The calculator prices it five ways: the subscription sticker, OpenAI and DeepSeek API rates, GPUs you buy, and GPUs you rent.

Open the calculator @@ -71,62 +73,24 @@ export default function Home() { ({ + ...relatedProduct(id), + relationship: id === "xcb" ? registeredRelationship("aicharts", id) : null, + })), }, { - heading: "The personal apps", + heading: productMessaging.headings["home-personal-group"], headingId: "related-apps", - items: [ - { - name: "PeopleBlade", - href: "https://peopleblade.com", - role: "Local personal CRM for everyone you know, built for your agent", - relationship: null, - }, - { - name: "Soulscrape", - href: "https://soulscrape.com", - role: "Free agent skill that writes dated dossiers on people, sources cited", - relationship: null, - }, - { - name: "Textbutler", - href: "https://textbutler.app", - role: "AI butler for the iMessage, WhatsApp, and Beeper chats you choose", - relationship: null, - }, - { - name: "Wordcell", - href: "https://wordcell.io", - role: "Markdown knowledge base that gives agents the decisions behind code", - relationship: null, - }, - ], + items: ["peopleblade", "soulscrape", "message-like-me", "kb"].map((id) => ({ + ...relatedProduct(id), + relationship: null, + })), }, ]} - heading="From the same workshop." + heading={productMessaging.headings["home-related"]} headingId="related-title" label="Related" summary="Other Hraness tools for people who work with AI agents." diff --git a/app/site.test.ts b/app/site.test.ts index 58c0c405..7572253e 100644 --- a/app/site.test.ts +++ b/app/site.test.ts @@ -1,3 +1,4 @@ +import { productMessaging } from "./messaging"; import { describe, expect, test } from "bun:test"; import packageJson from "../package.json"; @@ -16,19 +17,20 @@ import { describe("aicharts public positioning", () => { test("carries the canonical product messaging", () => { - // Canonical lines come from the portfolio messaging record (hraness/jungle - // a9988b903). The search title leads with the product name and names the - // comparison a searcher looks for; the tagline stays the visible heading. + // Search and visible copy share the checked portfolio snapshot. expect(searchSite.title.startsWith(`${site.name}: `)).toBeTrue(); for (const fact of ["benchmark scores", "cost per task"]) { expect(searchSite.title).toContain(fact); } expect(searchSite.title.length).toBeLessThanOrEqual(60); expect(searchSite.title).not.toContain("&"); - expect(site.tagline).toBe("See which model wins at each price."); - expect(site.category).toBe("AI model comparison charts"); - expect(homeHeading).toBe(site.tagline); + expect(site.tagline).toBe(productMessaging.tagline); + expect(site.category).toBe(productMessaging.category); + expect(homeHeading).toBe(productMessaging.hero.heading); expect(homeEyebrow).toBe(site.category); + expect(site.name).toBe(productMessaging.names.name); + expect(site.description).toBe(productMessaging.meta); + expect(homeLede).toBe(productMessaging.hero.summary); // The meta description names the two product halves: published benchmark // charts and the local token collector. expect(site.description.startsWith(`${site.name} `)).toBeTrue(); diff --git a/app/site.ts b/app/site.ts index df4a6241..1d0ec79a 100644 --- a/app/site.ts +++ b/app/site.ts @@ -1,18 +1,18 @@ -// Canonical public identity lines come from the portfolio messaging record -// for `aicharts` (hraness/jungle registry, commit a9988b903). +import { productMessaging, productName, productCanonicalUrl } from "./messaging"; + export const site = { - category: "AI model comparison charts", - description: "aicharts plots published AI benchmark scores against cost and tokens per task, and a local collector measures your own agents' token use.", + category: productMessaging.category, + description: productMessaging.meta, domain: "aicharts.io", emoji: "◉", - introduction: "Picking a model means guessing at a tradeoff between quality and price, because benchmark scores and prices live in different places. aicharts plots published benchmark scores against cost and tokens per task on one chart, so the strongest option at each budget is visible instead of implied. A local collector measures your own agents' token use, so the cost question covers your work, not only the models. aicharts is free and open source.", - name: "aicharts", - origin: "https://aicharts.io", + introduction: productMessaging.long, + name: productName, + origin: productCanonicalUrl, palette: { chromatic: { key: "#5e2e02", support: "#fefefd" }, tonal: { highlight: "#e1e0e0", shadow: "#291201" }, }, - tagline: "See which model wins at each price.", + tagline: productMessaging.tagline, } as const; export const searchSite = { @@ -20,18 +20,17 @@ export const searchSite = { name: site.name, origin: site.origin, socialImage: { - alt: "aicharts: Model benchmark scores plotted against cost and tokens per task", + alt: `${productName}: ${productMessaging.short}`, path: "/opengraph-image", }, - title: "aicharts: AI model benchmark scores vs cost per task", + title: productMessaging.headings["home-search-title"], } as const; export const homeEyebrow = site.category; -export const homeHeading = site.tagline; -export const homeLede = - "Benchmark scores plotted against cost and tokens per task, plus a local collector for your own agents' token use."; -export const homePrimaryAction = { href: "/#intelligence-index", label: "Browse the charts" } as const; -export const homeSecondaryAction = { href: "/usage", label: "Measure your agent" } as const; +export const homeHeading = productMessaging.hero.heading; +export const homeLede = productMessaging.hero.summary; +export const homePrimaryAction = { href: "/#intelligence-index", label: productMessaging.hero.primaryAction } as const; +export const homeSecondaryAction = { href: "/usage", label: productMessaging.hero.secondaryAction } as const; export const homeTaskLinks = [ { task: "coding", name: "Coding", description: "Build, debug, and work in a terminal." }, { task: "reasoning", name: "Reasoning", description: "Solve unfamiliar problems." }, @@ -42,7 +41,7 @@ export const homeTaskLinks = [ ] as const; export { usageReleaseUrl } from "@/lib/usage-cli-release"; -export const homeAboutHeading = "About aicharts"; +export const homeAboutHeading = productMessaging.headings["home-about"]; /** The date the comparison facts below were last checked against each product's own pages. */ export const homeAlternativesCheckedOn = "2026-09-28"; export const homeAlternativesLead = "Other places to compare models:"; diff --git a/app/social-image-site.ts b/app/social-image-site.ts index d36158cd..65c3eec5 100644 --- a/app/social-image-site.ts +++ b/app/social-image-site.ts @@ -7,6 +7,7 @@ import { import { formatIntelligenceCost, formatIntelligenceIndex } from "@/lib/index-model-pages"; import { MODEL_CARD_FALLBACK_CREATOR_SLUG } from "@/lib/model-card-route-status"; +import { productMessaging } from "./messaging"; import { aichartsMarkDataUrl } from "./social-image-mark"; import { modelCardsEyebrow, site } from "./site"; @@ -16,7 +17,7 @@ import { modelCardsEyebrow, site } from "./site"; * routes add page copy only. */ export const aichartsSocialImageSite = defineSocialImageSite({ - description: "Model benchmark scores plotted against cost and tokens per task.", + description: productMessaging.short, domain: site.domain, icon: { kind: "mark", src: aichartsMarkDataUrl }, // Names the cards set that must not break across lines. diff --git a/portfolio-messaging.generated.json b/portfolio-messaging.generated.json new file mode 100644 index 00000000..983cb750 --- /dev/null +++ b/portfolio-messaging.generated.json @@ -0,0 +1,1477 @@ +{ + "contract": "hraness.product-messaging/v1", + "productId": "aicharts", + "canonicalUrl": "https://aicharts.io", + "portfolio": { + "contract": "hraness.portfolio-public/v1", + "url": "https://hraness.com/portfolio.json", + "digest": "sha256:3e82dac495694ca14dbdea88c58d6781f23725a4005231eb0b5eb4e68d4e4907" + }, + "messaging": { + "formatVersion": 1, + "product": "aicharts", + "names": { + "name": "aicharts", + "command": "aicharts", + "formerly": [ + "CodingChart" + ] + }, + "category": "AI model comparison charts", + "tagline": "See which model wins at each price.", + "short": "Model benchmark scores plotted against cost and tokens per task", + "meta": "aicharts plots published AI benchmark scores against cost and tokens per task, and a local collector measures your own agents' token use.", + "medium": "aicharts plots published AI benchmark scores against cost and tokens per task, marking the best score at every budget. A local collector measures your own agents' token use.", + "long": "Picking a model means guessing at a tradeoff between quality and price, because benchmark scores and prices live in different places. aicharts plots published benchmark scores against cost and tokens per task on one chart, so the strongest option at each budget is visible instead of implied. A local collector measures your own agents' token use, so the cost question covers your work, not only the models. aicharts is free and open source.", + "hero": { + "heading": "See which model wins at each price.", + "summary": "Benchmark scores plotted against cost and tokens per task, plus a local collector for your own agents' token use.", + "primaryAction": "Browse the charts", + "secondaryAction": "Measure your agent" + }, + "headings": { + "home-calculator": "Subscription vs API vs GPUs", + "home-platform-group": "The agent platform", + "home-personal-group": "The personal apps", + "home-related": "From the same workshop.", + "home-about": "About aicharts", + "home-search-title": "aicharts: AI model benchmark scores vs cost per task" + }, + "status": { + "default": "proposed" + }, + "reviewedOn": "2026-09-24" + }, + "projects": [ + { + "id": "gobstopper", + "name": "Gobstopper", + "description": "compacts long agent sessions into smaller copies, keeping every byte", + "canonicalUrl": "https://gobstopper.sh", + "surface": { + "actionLabel": "explore Gobstopper", + "interfaces": [ + "CLI", + "MCP" + ], + "proof": "automatic context compaction that snapshots each transcript before changing it" + }, + "messaging": { + "names": { + "name": "Gobstopper", + "command": "gobstopper" + }, + "short": "Compacts long agent sessions into smaller copies, keeping every byte", + "tagline": "Context compaction you can undo." + } + }, + { + "id": "xcb", + "name": "Excalibur (xcb)", + "description": "routes coding tasks across the claude, codex, and devin plans you have", + "canonicalUrl": "https://xcb.sh", + "surface": { + "actionLabel": "explore Excalibur (xcb)", + "interfaces": [ + "CLI", + "SDK" + ], + "proof": "a customizable terminal workspace, provider account management, and token usage tracking" + }, + "messaging": { + "names": { + "name": "Excalibur (xcb)", + "command": "xcb", + "formerly": [ + "AgentMixer", + "OOMPA", + "HRA" + ] + }, + "short": "Routes coding tasks across the Claude, Codex, and Devin plans you have", + "tagline": "Use your Claude, Codex, and Devin plans from one agent." + } + }, + { + "id": "aicharts", + "name": "aicharts", + "description": "model benchmark scores plotted against cost and tokens per task", + "canonicalUrl": "https://aicharts.io", + "surface": { + "actionLabel": "compare models", + "interfaces": [ + "web app" + ], + "proof": "interactive Pareto frontier and model benchmark charts" + }, + "messaging": { + "names": { + "name": "aicharts", + "command": "aicharts", + "formerly": [ + "CodingChart" + ] + }, + "short": "Model benchmark scores plotted against cost and tokens per task", + "tagline": "See which model wins at each price." + } + }, + { + "id": "kb", + "name": "Wordcell", + "description": "markdown knowledge base that gives agents the decisions behind code", + "canonicalUrl": "https://wordcell.io", + "surface": { + "actionLabel": "open Wordcell", + "interfaces": [ + "CLI", + "SDK", + "Agent Skill" + ], + "proof": "Markdown recovered through backlinks, search, and Git context" + }, + "messaging": { + "names": { + "name": "Wordcell", + "command": "wordcell", + "formerly": [ + "KB" + ] + }, + "short": "Markdown knowledge base that gives agents the decisions behind code", + "tagline": "Give the next session what this one learned." + } + }, + { + "id": "oh-computer", + "name": "Oh", + "description": "memory for agents that stores each fact with its sources and history", + "canonicalUrl": "https://oh.computer", + "surface": { + "actionLabel": "research with Oh", + "interfaces": [ + "CLI", + "SDK", + "Agent Skill" + ], + "proof": "an inspectable question-to-citation research trace" + }, + "messaging": { + "names": { + "name": "Oh", + "command": "oh" + }, + "short": "Memory for agents that stores each fact with its sources and history", + "tagline": "Agent memory that shows its work." + } + }, + { + "id": "sponge", + "name": "Sponge", + "description": "local research service that keeps its sources and writes cited reports", + "canonicalUrl": "https://sponge.computer", + "surface": { + "actionLabel": "Explore Sponge", + "interfaces": [ + "CLI", + "terminal UI" + ], + "proof": "cited reports from saved sources, with strategy changes tested in a sealed evaluation before you adopt them" + }, + "messaging": { + "names": { + "name": "Sponge", + "command": "spongev2" + }, + "short": "Local research service that keeps its sources and writes cited reports", + "tagline": "Deep research that runs on your own machine." + } + }, + { + "id": "algal", + "name": "ALGAL", + "description": "language and vm for agent programs that wait for approval and resume", + "canonicalUrl": "https://algal.computer", + "surface": { + "actionLabel": "explore ALGAL", + "interfaces": [ + "CLI", + "SDK" + ], + "proof": "content-addressed programs, capabilities with declared limits, and replayable run receipts" + }, + "messaging": { + "names": { + "name": "ALGAL", + "command": "algal", + "formerly": [ + "Morphogen" + ] + }, + "short": "Language and VM for agent programs that wait for approval and resume", + "tagline": "Write agent programs that wait, resume, and replay." + } + }, + { + "id": "valhalla", + "name": "Valhalla", + "description": "peer-to-peer rooms where agents and their owners share signed work", + "canonicalUrl": "https://vhalla.com", + "surface": { + "actionLabel": "explore Valhalla", + "interfaces": [ + "CLI", + "web app" + ], + "proof": "an early Rust project for signed agent chat and browser participation" + }, + "messaging": { + "names": { + "name": "Valhalla", + "command": "vhalla" + }, + "short": "Peer-to-peer rooms where agents and their owners share signed work", + "tagline": "A meeting place for agents, run by the people in it." + } + }, + { + "id": "peopleblade", + "name": "PeopleBlade", + "description": "the contact book you own, built for your agent", + "canonicalUrl": "https://peopleblade.com", + "surface": { + "actionLabel": "open PeopleBlade", + "interfaces": [ + "CLI", + "Agent Skill" + ], + "proof": "your contacts in one local database, each record kept with its source" + }, + "messaging": { + "names": { + "name": "PeopleBlade", + "command": "peopleblade" + }, + "short": "The contact book you own, built for your agent", + "tagline": "A local personal CRM for everyone you know." + } + }, + { + "id": "message-like-me", + "name": "TextButler", + "description": "ai butler for the imessage, whatsapp, and beeper chats you choose", + "canonicalUrl": "https://textbutler.app", + "surface": { + "actionLabel": "open TextButler", + "interfaces": [ + "CLI" + ], + "proof": "marked replies in the one-to-one chats you turn on, with notes kept on your Mac" + }, + "messaging": { + "names": { + "name": "TextButler", + "command": "textbutler", + "formerly": [ + "Message Like Me" + ] + }, + "short": "AI butler for the iMessage, WhatsApp, and Beeper chats you choose", + "tagline": "An AI butler in your messaging apps." + } + }, + { + "id": "wrench", + "name": "GhostGet", + "description": "named web actions for ai agents: read pages, save media, use connected accounts", + "canonicalUrl": "https://ghostget.com", + "surface": { + "actionLabel": "open GhostGet", + "interfaces": [ + "CLI", + "SDK", + "Agent Skill" + ], + "proof": "signed web captures and provider actions within declared limits" + }, + "messaging": { + "names": { + "name": "GhostGet", + "command": "ghostget", + "formerly": [ + "Wrench" + ] + }, + "short": "Named web actions for AI agents: read pages, save media, use connected accounts", + "tagline": "Your agent calls web actions by name and holds no password." + } + }, + { + "id": "soulscrape", + "name": "Soulscrape", + "description": "free agent skill that writes dated dossiers on people, sources cited", + "canonicalUrl": "https://soulscrape.com", + "surface": { + "actionLabel": "open Soulscrape", + "interfaces": [ + "Agent Skill" + ], + "proof": "a dated person model with cited evidence and open questions" + }, + "messaging": { + "names": { + "name": "Soulscrape", + "command": "@hraness/soulscrape", + "formerly": [ + "Ensoul" + ] + }, + "short": "Free agent skill that writes dated dossiers on people, sources cited", + "tagline": "See how someone thinks, and where every claim comes from." + } + }, + { + "id": "iconplace", + "name": "icon.place", + "description": "editable vector icons and illustrations, with an editor in your browser", + "canonicalUrl": "https://icon.place", + "surface": { + "actionLabel": "open icon.place", + "interfaces": [ + "web app" + ], + "proof": "9,518 concept pages and 8,965 drawn icons that export as SVG, PNG, or a recipe" + }, + "messaging": { + "names": { + "name": "icon.place" + }, + "short": "Editable vector icons and illustrations, with an editor in your browser", + "tagline": "Icons and illustrations you can edit as vectors." + } + }, + { + "id": "slopcamera", + "name": "SlopCamera", + "description": "a multimedia studio for your coding agent", + "canonicalUrl": "https://slopcamera.com", + "surface": { + "actionLabel": "open SlopCamera", + "interfaces": [ + "CLI", + "SDK", + "MCP", + "Agent Skill" + ], + "proof": "generated media with operation records that identify inputs and outputs" + }, + "messaging": { + "names": { + "name": "SlopCamera", + "command": "slopcamera", + "formerly": [ + "Atet", + "Graphics" + ] + }, + "short": "A multimedia studio for your coding agent", + "tagline": "A multimedia studio for your coding agent." + } + }, + { + "id": "soundfish", + "name": "Soundfish", + "description": "album pages with a waveform player and comments pinned to moments", + "canonicalUrl": "https://sound.fish", + "surface": { + "actionLabel": "open the editor", + "interfaces": [ + "web app", + "CLI", + "Agent Skill" + ], + "proof": "a MIDI composition encoded in a content-addressed URL" + }, + "messaging": { + "names": { + "name": "Soundfish", + "command": "soundfish" + }, + "short": "Album pages with a waveform player and comments pinned to moments", + "tagline": "Give your album a page of its own." + } + }, + { + "id": "sys1", + "name": "Sys1", + "description": "saves repository check results and adds scoped code review", + "canonicalUrl": "https://sys1.io", + "surface": { + "actionLabel": "explore Sys1", + "interfaces": [ + "CLI", + "SDK" + ], + "proof": "saved repository checks, scoped code review, and structured model answers through a CLI and SDK" + }, + "messaging": { + "names": { + "name": "Sys1", + "command": "sys1" + }, + "short": "Saves repository check results and adds scoped code review", + "tagline": "Saved checks and code review" + } + }, + { + "id": "sloptrade", + "name": "SlopTrade", + "description": "build prompt that turns a coding agent into a trading-system designer", + "canonicalUrl": "https://sloptrade.com", + "surface": { + "actionLabel": "explore Sloptrade", + "interfaces": [ + "website", + "publication" + ], + "proof": "a citation-backed architecture dossier with system and research notes" + }, + "messaging": { + "names": { + "name": "SlopTrade", + "command": "sloptrade", + "formerly": [ + "Invest" + ] + }, + "short": "Build prompt that turns a coding agent into a trading-system designer", + "tagline": "Design an AI trading system that fails closed." + } + }, + { + "id": "hraness", + "name": "Hraness", + "description": "software studio making tools for ai agents and for people", + "canonicalUrl": "https://hraness.com", + "surface": { + "actionLabel": "open Hraness", + "interfaces": [ + "website" + ], + "proof": "organization statement with project, writing, reference, and Reading indexes" + }, + "messaging": { + "names": { + "name": "Hraness" + }, + "short": "Software studio making tools for AI agents and for people", + "tagline": "Tools for agents and humans." + } + }, + { + "id": "sleepyland", + "name": "Sleepyland", + "description": "free sleep sounds made in your browser, with sourced sleep guides", + "canonicalUrl": "https://sleepy.land", + "surface": { + "actionLabel": "explore Sleepyland", + "interfaces": [ + "web app", + "publication" + ], + "proof": "a browser sound machine beside cited sleep research" + }, + "messaging": { + "names": { + "name": "Sleepyland" + }, + "short": "Free sleep sounds made in your browser, with sourced sleep guides", + "tagline": "Steady sound for sleep, made in your browser." + } + }, + { + "id": "roughday", + "name": "Rough Day", + "description": "daily front page that says why each story ranked and links its source", + "canonicalUrl": "https://rough.day", + "surface": { + "actionLabel": "read today's archive", + "interfaces": [ + "archive" + ], + "proof": "a dated, source-linked daily news archive" + }, + "messaging": { + "names": { + "name": "Rough Day" + }, + "short": "Daily front page that says why each story ranked and links its source", + "tagline": "A daily front page that says why each story ranked." + } + }, + { + "id": "stripe-history", + "name": "Stripe History", + "description": "independent history of stripe where every event is dated and sourced", + "canonicalUrl": "https://hraness.com/stripe", + "surface": { + "actionLabel": "explore the timeline", + "interfaces": [ + "archive" + ], + "proof": "a source-linked company timeline and charts" + }, + "messaging": { + "names": { + "name": "Stripe History", + "formerly": [ + "Stripe Guide", + "Stripedex" + ] + }, + "short": "Independent history of Stripe where every event is dated and sourced", + "tagline": "Every event in Stripe's history, dated and sourced." + } + }, + { + "id": "eds-research", + "name": "EDS Research Index", + "description": "ehlers-danlos evidence stratified by kind, every record linked to sources", + "canonicalUrl": "https://hraness.com/eds", + "surface": { + "actionLabel": "explore the index", + "interfaces": [ + "archive" + ], + "proof": "an Ehlers-Danlos research index where every record names its sources and evidence tier" + }, + "messaging": { + "names": { + "name": "EDS Research Index" + }, + "short": "Ehlers-Danlos evidence stratified by kind, every record linked to sources", + "tagline": "Ehlers-Danlos research, sorted by kind of evidence." + } + }, + { + "id": "act60", + "name": "act60.me", + "description": "estimate act 60 savings after real costs, with sourced guides and a day tracker", + "canonicalUrl": "https://act60.me", + "surface": { + "actionLabel": "estimate Act 60 savings", + "interfaces": [ + "web app" + ], + "proof": "an Act 60 savings calculator that shows its assumptions and the date the law was last checked" + }, + "messaging": { + "names": { + "name": "act60.me", + "shortName": "ACT60" + }, + "short": "Estimate Act 60 savings after real costs, with sourced guides and a day tracker", + "tagline": "Model the move to Puerto Rico before you make it." + } + }, + { + "id": "direct", + "name": "Direct", + "description": "repeatable app states for browser agents, opened by url", + "canonicalUrl": "https://hraness.com/direct", + "surface": { + "actionLabel": "open Direct", + "interfaces": [ + "library", + "Agent Skill" + ], + "proof": "a fixture, scenario, and deterministic browser trace" + }, + "messaging": { + "names": { + "name": "Direct", + "command": "@hraness/direct" + }, + "short": "Repeatable app states for browser agents, opened by URL", + "tagline": "Give each app state you test its own URL." + } + }, + { + "id": "clankdar", + "name": "Clankdar", + "description": "fresh puzzles for ai agents, scored exactly, with a signed receipt", + "canonicalUrl": "https://clankdar.com", + "surface": { + "actionLabel": "explore Clankdar", + "interfaces": [ + "website" + ], + "proof": "reproducible capability checks with verifiable results" + }, + "messaging": { + "names": { + "name": "Clankdar", + "command": "clankdar", + "formerly": [ + "botcaptcha" + ] + }, + "short": "Fresh puzzles for AI agents, scored exactly, with a signed receipt", + "tagline": "Check what your agent can solve." + } + }, + { + "id": "lifedaysleft", + "name": "Lifecharts", + "description": "turn the chapters of your life into one timeline you can share", + "canonicalUrl": "https://lifecharts.io", + "surface": { + "actionLabel": "make a timeline", + "interfaces": [ + "web app", + "CLI" + ], + "proof": "editable life chapters with shareable links and website embeds" + }, + "messaging": { + "names": { + "name": "Lifecharts", + "command": "lifecharts", + "formerly": [ + "Life Days Left", + "lifedaysleft" + ] + }, + "short": "Turn the chapters of your life into one timeline you can share", + "tagline": "Chart your life in chapters." + } + }, + { + "id": "swft", + "name": "SWFT", + "description": "free publication about how companies put ai agents to work", + "canonicalUrl": "https://swft.io", + "surface": { + "actionLabel": "read SWFT", + "interfaces": [ + "publication" + ], + "proof": "reported software-factory stories with source notes" + }, + "messaging": { + "names": { + "name": "SWFT" + }, + "short": "Free publication about how companies put AI agents to work", + "tagline": "How companies put AI agents to work." + } + }, + { + "id": "pattern-language", + "name": "Pattern Language", + "description": "tests whether alexander-inspired patterns help agents write better code", + "canonicalUrl": "https://github.com/hraness/pattern-language", + "surface": { + "actionLabel": "read the studies", + "interfaces": [ + "archive" + ], + "proof": "executable code-design studies that record when pattern guidance helps and when it does not" + }, + "messaging": { + "names": { + "name": "Pattern Language" + }, + "short": "Tests whether Alexander-inspired patterns help agents write better code", + "tagline": "Alexander's notes, sketched as runnable ALGAL form." + } + }, + { + "id": "system-one-skills", + "name": "System One Skills", + "description": "runs a check once, returns a short result, and keeps the full log", + "canonicalUrl": "https://sys1.io/skills", + "surface": { + "actionLabel": "install the skill", + "interfaces": [ + "Agent Skill", + "CLI" + ], + "proof": "one deterministic skill that cut validation-log text 35.20 percent across 563 replayed outputs" + }, + "messaging": { + "names": { + "name": "System One Skills", + "command": "system-one-skills" + }, + "short": "Runs a check once, returns a short result, and keeps the full log", + "tagline": "Cut noisy validation logs without a model call." + } + }, + { + "id": "ghostget-skills", + "name": "GhostGet Skills", + "description": "packages provider reads and plan checks as programs with no model calls", + "canonicalUrl": "https://github.com/hraness/ghostget-skills", + "surface": { + "actionLabel": "install the skills", + "interfaces": [ + "Agent Skill", + "CLI" + ], + "proof": "ALGAL programs for GhostGet reads, plan checks, and drift watch, with zero model calls" + }, + "messaging": { + "names": { + "name": "GhostGet Skills", + "command": "ghostget-skills" + }, + "short": "Packages provider reads and plan checks as programs with no model calls", + "tagline": "GhostGet workflows as programs you can replay." + } + } + ], + "relations": [ + { + "source": "hraness", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "hraness", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "hraness", + "target": "posthog-sdk", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "hraness", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "hraness", + "target": "web-discovery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "aicharts", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "aicharts", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "aicharts", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "aicharts", + "target": "web-discovery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "swft", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "swft", + "target": "web-discovery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "lifedaysleft", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "lifedaysleft", + "target": "result", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "lifedaysleft", + "target": "slopcamera", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sleepyland", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sleepyland", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sleepyland", + "target": "posthog-sdk", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sleepyland", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sleepyland", + "target": "web-discovery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "roughday", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "roughday", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "roughday", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "roughday", + "target": "direct", + "kind": "development", + "label": "develops with", + "direction": "forward" + }, + { + "source": "stripe-history", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "stripe-history", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "stripe-history", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "stripe-history", + "target": "web-discovery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "stripe-history", + "target": "hraness", + "kind": "delivery", + "label": "published at /stripe", + "direction": "forward", + "detail": "Hraness proxies the standalone Stripe History data application under its canonical /stripe route." + }, + { + "source": "eds-research", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "eds-research", + "target": "web-discovery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "eds-research", + "target": "hraness", + "kind": "delivery", + "label": "published at /eds", + "direction": "forward", + "detail": "Hraness proxies the standalone EDS research index under its canonical /eds route." + }, + { + "source": "act60", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "act60", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "act60", + "target": "result", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "act60", + "target": "convex-toolkit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "act60", + "target": "suite-accounts", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "act60", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "wrench", + "target": "kb", + "kind": "runtime", + "label": "ships KB workflows", + "direction": "forward" + }, + { + "source": "wrench", + "target": "message-like-me", + "kind": "contract", + "label": "exports private bundles", + "direction": "forward", + "detail": "GhostGet produces bounded multi-account Beeper bundles that TextButler verifies and ingests locally." + }, + { + "source": "peopleblade", + "target": "wrench", + "kind": "runtime", + "label": "provider transport", + "direction": "forward", + "detail": "PeopleBlade uses GhostGet to read contacts and search results from connected accounts such as Beeper and WhatsApp, and checks each result before saving it locally." + }, + { + "source": "message-like-me", + "target": "peopleblade", + "kind": "contract", + "label": "shared bundle format", + "direction": "shared", + "detail": "Both tools implement the versioned message-like-me.local-message-bundle interchange without importing each other's private state." + }, + { + "source": "gobstopper", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "gobstopper", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "gobstopper", + "target": "xcb", + "kind": "runtime", + "label": "compacts sessions for", + "direction": "forward", + "detail": "Excalibur (xcb) uses Gobstopper's elision policy to drop stale tool output from Claude Code and Codex prompts once context passes a threshold, and keeps the original output in local history. It is on by default." + }, + { + "source": "xcb", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "xcb", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "xcb", + "target": "aicharts", + "kind": "contract", + "label": "exports sessions for", + "direction": "forward", + "detail": "Excalibur (xcb) measures subscription usage locally and, when you turn on exports, writes session files in the aicharts format. Automatic upload is not available." + }, + { + "source": "soulscrape", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "soulscrape", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "soulscrape", + "target": "peopleblade", + "kind": "contract", + "label": "exports dossier packets", + "direction": "forward", + "detail": "Soulscrape exports dated dossier packets that PeopleBlade imports and rebinds to contacts during local merge review." + }, + { + "source": "wrench", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "wrench", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "message-like-me", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "message-like-me", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "slopcamera", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "slopcamera", + "target": "direct", + "kind": "development", + "label": "develops with", + "direction": "forward" + }, + { + "source": "peopleblade", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "peopleblade", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "peopleblade", + "target": "posthog-sdk", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sponge", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sponge", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sponge", + "target": "convex-toolkit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sponge", + "target": "posthog-sdk", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sponge", + "target": "suite-accounts", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sponge", + "target": "vercel-delivery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "sponge", + "target": "web-discovery", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "kb", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "kb", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "direct", + "target": "hraness", + "kind": "delivery", + "label": "README-backed /direct", + "direction": "forward", + "detail": "Hraness renders the pinned public Direct README and Agent Skill at /direct." + }, + { + "source": "soundfish", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "soundfish", + "target": "result", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "soundfish", + "target": "direct", + "kind": "development", + "label": "develops with", + "direction": "forward" + }, + { + "source": "oh-computer", + "target": "ui", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "oh-computer", + "target": "design-kit", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "oh-computer", + "target": "oh", + "kind": "delivery", + "label": "publishes kernel specification", + "direction": "forward", + "detail": "The public Oh site publishes the versioned specification for the @hraness/oh kernel from the same repository." + }, + { + "source": "design-kit", + "target": "ui", + "kind": "runtime", + "label": "builds on", + "direction": "forward" + }, + { + "source": "suite-accounts", + "target": "result", + "kind": "runtime", + "label": "uses", + "direction": "forward" + }, + { + "source": "suite-accounts", + "target": "accounts", + "kind": "contract", + "label": "OAuth protocol", + "direction": "forward", + "detail": "The public Suite Accounts package validates the protocol spoken by products and the Accounts authority." + }, + { + "source": "message-like-me", + "target": "xcb", + "kind": "runtime", + "label": "drafts replies through", + "direction": "forward", + "detail": "TextButler can classify messages and draft replies through Excalibur (xcb) on the Claude Code or Codex subscription you already pay for, with xcb holding the sign-in; a separately billed API route also exists." + }, + { + "source": "message-like-me", + "target": "algal", + "kind": "runtime", + "label": "runs reply habitats on", + "direction": "forward", + "detail": "TextButler's opt-in per-contact habitats run as ALGAL programs: a candidate reply plan replaces the current one only after a blind replay shows no regression, and no plan can change the recipient, provider, or permissions." + }, + { + "source": "sponge", + "target": "oh-computer", + "kind": "runtime", + "label": "keeps its hosted library's agent working memory in", + "direction": "forward", + "detail": "Sponge's hosted library at sponge.computer keeps its research agents' working memory in a server-side Oh store, separate from the reviewed knowledge in its product database." + }, + { + "source": "kb", + "target": "oh-computer", + "kind": "runtime", + "label": "answers graph queries with", + "direction": "forward", + "detail": "Wordcell rebuilds a disposable Oh graph from your Markdown to answer named graph queries with source proofs; Markdown and Git stay the record, and search does not use Oh's memory retrieval." + }, + { + "source": "xcb", + "target": "algal", + "kind": "runtime", + "label": "replays task history with", + "direction": "forward", + "detail": "Excalibur (xcb) embeds the ALGAL runtime for task transitions, route and settle reflexes, and resumable controllers, and its task verifier replays a task's recorded history offline; the managed harness is experimental." + }, + { + "source": "xcb", + "target": "kb", + "kind": "runtime", + "label": "searches project notes with", + "direction": "forward", + "detail": "Excalibur (xcb) can bind one explicit Wordcell vault through a hash-pinned CLI, giving workers a read-only memory search with citations; saving a note back to the vault is always an explicit step." + }, + { + "source": "clankdar", + "target": "algal", + "kind": "runtime", + "label": "scores puzzles with", + "direction": "forward", + "detail": "Clankdar's default puzzles are small programs in ALGAL's expression language, and ALGAL's official evaluator, pinned by commit and hash, computes each reference answer, so no judge model decides." + }, + { + "source": "slopcamera", + "target": "algal", + "kind": "runtime", + "label": "bakes character behavior with", + "direction": "forward", + "detail": "SlopCamera's scene behavior bake runs character behavior as ALGAL organisms with no executors or side effects." + }, + { + "source": "sponge", + "target": "wrench", + "kind": "runtime", + "label": "captures signed-in pages through", + "direction": "forward", + "detail": "Sponge captures signed-in pages through GhostGet, storing only a login name, and the hosted library at sponge.computer imports GhostGet's text-only capture bundles offline." + }, + { + "source": "sponge", + "target": "soulscrape", + "kind": "contract", + "label": "imports research exchanges into its hosted library from", + "direction": "forward", + "detail": "Sponge's hosted library at sponge.computer converts a public Soulscrape research exchange offline into an Oh research packet that it does not treat as established fact." + }, + { + "source": "xcb", + "target": "relay", + "kind": "runtime", + "label": "uses", + "direction": "forward", + "detail": "Excalibur (xcb) depends on @hraness/relay and instantiates its Convex relay modules for devices, commands, envelopes, invites, projections, and maintenance." + }, + { + "source": "pattern-language", + "target": "algal", + "kind": "development", + "label": "runs studies with", + "direction": "forward", + "detail": "Pattern Language uses ALGAL so its design experiments stay inspectable: content-addressed manifests and records you can replay." + }, + { + "source": "system-one-skills", + "target": "sys1", + "kind": "delivery", + "label": "skills guide hosted by", + "direction": "forward", + "detail": "The System One Skills guide is published at sys1.io/skills on the Sys1 site; the skill pack does not require Sys1 at runtime." + }, + { + "source": "ghostget-skills", + "target": "wrench", + "kind": "runtime", + "label": "uses", + "direction": "forward", + "detail": "GhostGet Skills wraps the GhostGet CLI for catalog, plan check, invoke, read, doctor, and auth list, and pins @hraness/ghostget." + }, + { + "source": "ghostget-skills", + "target": "algal", + "kind": "runtime", + "label": "runs programs on", + "direction": "forward", + "detail": "Each GhostGet Skills workflow is an ALGAL program the ALGAL runtime executes; the package pins @hraness/algal." + }, + { + "source": "message-like-me", + "target": "wrench", + "kind": "runtime", + "label": "reads and sends messages through", + "direction": "forward", + "detail": "Messages reach TextButler through GhostGet, a separate Mac tool you install first for messaging sign-in and permissions, and TextButler's marked replies go back the same way." + }, + { + "source": "sponge", + "target": "kb", + "kind": "runtime", + "label": "keeps documents in", + "direction": "forward", + "detail": "Sponge uses Wordcell for documents and PDF text, renders a static Wordcell edition when you publish, and can put selected documents on a wordcell.io site." + }, + { + "source": "sponge", + "target": "algal", + "kind": "runtime", + "label": "evolves research strategies with", + "direction": "forward", + "detail": "Sponge runs on ALGAL, and a candidate strategy for brief research must pass a sealed evaluation that ALGAL replays; nothing changes until you adopt it or restore the previous one." + }, + { + "source": "sponge", + "target": "xcb", + "kind": "runtime", + "label": "reaches models through", + "direction": "forward", + "detail": "Sponge reaches every model through a hash-pinned Excalibur (xcb) executable, so provider sign-ins stay in xcb." + }, + { + "source": "algal", + "target": "xcb", + "kind": "runtime", + "label": "delegates coding tasks to", + "direction": "forward", + "detail": "ALGAL's xcb executor hands coding tasks to Excalibur (xcb) through its JSON run command, and ALGAL never copies credentials or reimplements account switching." + }, + { + "source": "soundfish", + "target": "algal", + "kind": "runtime", + "label": "evolves loops with", + "direction": "forward", + "detail": "Soundfish's evolve command turns a loop into an evolving recipe that a bundled ALGAL evaluator, pinned by SHA-256, runs offline." + }, + { + "source": "aicharts", + "target": "gobstopper", + "kind": "runtime", + "label": "reads compaction telemetry from", + "direction": "forward", + "detail": "The aicharts collector parses Gobstopper's numeric compaction telemetry and never reads transcript content." + }, + { + "source": "iconplace", + "target": "algal", + "kind": "runtime", + "label": "records drawing runs with", + "direction": "forward", + "detail": "The icon.place drawing lab runs each experiment as an ALGAL program in the browser and records the proposal, candidates, and selection as receipts it can replay; the live site does not use ALGAL yet." + } + ] +} diff --git a/scripts/model-card-browser.ts b/scripts/model-card-browser.ts index 794bc5ef..5a9e7df0 100644 --- a/scripts/model-card-browser.ts +++ b/scripts/model-card-browser.ts @@ -3,6 +3,7 @@ import { join } from "node:path"; import { ATLAS_DATASETS } from "../lib/benchmark-atlas-catalog"; import { verifyUsageDashboard } from "./usage-browser"; import { verifyUsageSessions } from "./usage-sessions-browser"; +import { ownedChromiumOptions } from "./owned-chromium.mjs"; import { chromium, @@ -37,11 +38,14 @@ async function launchFirstAvailableBrowser(paths: readonly string[]): Promise
typeof feature === "string" && /^[A-Za-z0-9]+$/u.test(feature)), "Expected constant pinned Playwright feature names"); + +/** Retain the pinned Playwright defaults and replace only its feature switch. */ +export function ownedChromiumOptions(executablePath) { + assert.ok(isAbsolute(executablePath), "Owned Chromium needs an absolute executable path"); + const executable = realpathSync(executablePath); + assert.ok(!executable.includes("/Google Chrome.app/"), "The installed Google Chrome app is not an owned test browser"); + const existing = [`--disable-features=${pinnedFeatures.join(",")}`]; + const features = new Set([...existing[0].slice("--disable-features=".length).split(","), "PaintHolding", "MacAppCodeSignClone"]); + return { + executablePath: executable, + ignoreDefaultArgs: existing, + args: ["--mute-audio", `--disable-features=${[...features].join(",")}`], + }; +} diff --git a/scripts/owned-chromium.test.ts b/scripts/owned-chromium.test.ts new file mode 100644 index 00000000..49947634 --- /dev/null +++ b/scripts/owned-chromium.test.ts @@ -0,0 +1,36 @@ +import { expect, test } from "bun:test"; +import { mkdtemp, mkdir, rm, symlink, writeFile } from "node:fs/promises"; +import { tmpdir } from "node:os"; +import { join } from "node:path"; +import { ownedChromiumOptions } from "./owned-chromium.mjs"; + +test("owned browser options preserve Playwright features in one muted launch switch", () => { + const options = ownedChromiumOptions(process.execPath); + expect(options.args).toContain("--mute-audio"); + const switches = options.args.filter((argument: string) => argument.startsWith("--disable-features=")); + expect(switches).toHaveLength(1); + const features = switches[0]!.slice("--disable-features=".length).split(","); + for (const feature of ["PaintHolding", "MacAppCodeSignClone", "MediaRouter", "ThirdPartyStoragePartitioning"]) { + expect(features).toContain(feature); + } + expect(new Set(features).size).toBe(features.length); + expect(options.ignoreDefaultArgs).toHaveLength(1); + expect(options.ignoreDefaultArgs[0]).toStartWith("--disable-features="); +}); + +test("rejects the installed Chrome app even through an executable symlink", async () => { + const root = await mkdtemp(join(tmpdir(), "aicharts-browser-selection-")); + try { + const app = join(root, "Google Chrome.app", "Contents", "MacOS"); + await mkdir(app, { recursive: true }); + const executable = join(app, "Google Chrome"); + await writeFile(executable, "inert test fixture"); + const alias = join(root, "browser"); + await symlink(executable, alias); + expect(() => ownedChromiumOptions(executable)).toThrow("not an owned test browser"); + expect(() => ownedChromiumOptions(alias)).toThrow("not an owned test browser"); + expect(() => ownedChromiumOptions("./browser")).toThrow("absolute executable path"); + } finally { + await rm(root, { recursive: true, force: true }); + } +}); diff --git a/scripts/verify-public-site-browser.mjs b/scripts/verify-public-site-browser.mjs index fb0d0d9e..6602d9a4 100644 --- a/scripts/verify-public-site-browser.mjs +++ b/scripts/verify-public-site-browser.mjs @@ -6,6 +6,7 @@ import { createServer } from "node:net"; import { resolve } from "node:path"; import { fileURLToPath } from "node:url"; import { chromium } from "playwright-core"; +import { ownedChromiumOptions } from "./owned-chromium.mjs"; const root = fileURLToPath(new URL("../", import.meta.url)); const production = process.argv.includes("--production"); @@ -49,7 +50,9 @@ try { } assert.ok(ready, `Next did not become ready: ${output}`); } - browser = await chromium.launch({ executablePath: await browserExecutable() }); + const owned = ownedChromiumOptions(await browserExecutable()); + browser = await chromium.launch(owned); + console.log(`Owned Chromium: ${owned.executablePath} (${browser.version()})`); // Contexts are isolated and share only the server, so a small pool of them runs at once. // Each combination records its own results; they are joined in the fixed combination order. const combinations = [360, 390, 1440].flatMap(width => ["light", "dark"].map(theme => ({ width, theme }))); @@ -80,15 +83,14 @@ await writeFile(resolve(artifacts, "results.json"), JSON.stringify({ passed: !fa if (failure) throw failure; console.log(`Verified ${results.length} route/viewport/theme combinations at ${origin}.`); -/** Prefer an explicit or preinstalled Chrome so CI needs no browser download. */ +/** Prefer an explicit test browser or the Chromium pinned by this app. */ async function browserExecutable() { - const candidates = [process.env.CHROMIUM_EXECUTABLE_PATH, chromium.executablePath(), "/usr/bin/google-chrome", "/usr/bin/chromium", "/usr/bin/chromium-browser", - "/Applications/Google Chrome.app/Contents/MacOS/Google Chrome", "/Applications/Chromium.app/Contents/MacOS/Chromium"]; + const candidates = [process.env.CHROMIUM_EXECUTABLE_PATH, chromium.executablePath()]; for (const candidate of candidates) { if (!candidate) continue; try { await access(candidate); return candidate; } catch { /* Try the next installed browser. */ } } - return undefined; + throw new Error("No pinned Chromium executable found; provision this app's Playwright browser."); } async function verifyCombination({ width, theme }, results) { diff --git a/services/usage-worker/test/contribution-projection.worker.ts b/services/usage-worker/test/contribution-projection.worker.ts index 89a6150a..b9e1039b 100644 --- a/services/usage-worker/test/contribution-projection.worker.ts +++ b/services/usage-worker/test/contribution-projection.worker.ts @@ -38,9 +38,10 @@ function transaction(storage: DurableObjectStorage, admitted: () => boolean): Ad } const observation = (): AdmissionObservation => ({ generation: env.USAGE_ENROLLMENT_GENERATION, observed: now, committed: false, fence: null }); const tick = (duration = CONTRIBUTION_PROJECTION_PUBLISH_INTERVAL_MS) => { now += duration; vi.setSystemTime(now); }; -async function advance(overrides: Partial = {}, admitted: () => boolean = () => true): Promise { - return runInDurableObject(stub(), (_instance, context) => new AccountContributionProjection({ ...env, ...overrides }, - new ContributionProjectionState(context.storage), transaction(context.storage, admitted)).advance({ schemaVersion: 3, ...scope() }, observation())); +async function advance(overrides: Partial = {}, admitted: () => boolean = () => true, storage?: DurableObjectStorage): Promise { + const run = (value: DurableObjectStorage) => new AccountContributionProjection({ ...env, ...overrides }, + new ContributionProjectionState(value), transaction(value, admitted)).advance({ schemaVersion: 3, ...scope() }, observation()); + return storage === undefined ? runInDurableObject(stub(), (_instance, context) => run(context.storage)) : run(storage); } function success(result: ContributionProjectionAdvanceResult): ContributionProjectionStatus { if (!result.ok) throw new Error(`synthetic projection refusal: ${result.error}`); return result.value; @@ -52,8 +53,8 @@ const row = (id: number, corrected = false): UsageStatsRow => { durationMs: null, timedRecords: 0, timedTokens: "0", tokenBasis: "reported", breakdownCoverage: "complete" }); if (!value) throw new Error("invalid synthetic row"); return value; }; -async function batch(count: number, corrected = false, removeAbove = Number.MAX_SAFE_INTEGER, deviceId = DEVICE): Promise { - return canonical(state => { +async function batch(count: number, corrected = false, removeAbove = Number.MAX_SAFE_INTEGER, deviceId = DEVICE, run: typeof canonical = canonical): Promise { + return run(state => { const population = state.population(POPULATION)!; const value = parseContributionBatch({ schemaVersion: 3, profile: CONTRIBUTION_PROFILE, identityScheme: CONTRIBUTION_IDENTITY, grain: "observation", ...scope(), deviceId, operationId: hex(++operation), sequence: state.sequence(env.USAGE_ENROLLMENT_GENERATION, deviceId) + 1, @@ -64,11 +65,11 @@ async function batch(count: number, corrected = false, removeAbove = Number.MAX_ if (!value) throw new Error("invalid synthetic batch"); return value; }); } -async function publish(request: ContributionBatch) { - await canonical(state => state.reserve(request, authority())); +async function publish(request: ContributionBatch, run: typeof canonical = canonical) { + await run(state => state.reserve(request, authority())); const body = await ensureContributionBody(env.STAGING, request, () => true); if (!body.ok) throw new Error(body.error); - const bundle = await canonical(state => state.deltaBundle(request, authority())), proof = await ensureContributionJournal(env.STAGING, bundle, () => true); - await canonical(state => state.commit(request, body.value, authority(), proof)); + const bundle = await run(state => state.deltaBundle(request, authority())), proof = await ensureContributionJournal(env.STAGING, bundle, () => true); + await run(state => state.commit(request, body.value, authority(), proof)); } async function grantEmpty(): Promise { await canonical(state => state.grantPopulation({ schemaVersion: 3, ...scope(), deviceId: DEVICE, @@ -82,8 +83,8 @@ async function converge(): Promise { } throw new Error("bounded synthetic convergence exceeded"); } -async function applyAll(): Promise { - for (let step = 0; step < 600; step++) { const result = success(await advance()); if (!result.appliedLag) return result; } +async function applyAll(storage?: DurableObjectStorage): Promise { + for (let step = 0; step < 600; step++) { const result = success(await advance({}, () => true, storage)); if (!result.appliedLag) return result; } throw new Error("bounded synthetic application exceeded"); } async function prepareNumeric(): Promise { @@ -278,24 +279,30 @@ test("more than 64 numeric revisions apply in one cursor horizon and coalesce to await publish(await batch(4)); await converge(); const prior = await cells(3), expected = new Map(Array.from({ length: 4 }, (_, index) => [hex(index + 1, 32), row(index + 1)])); const started = now; - for (let revision = 0; revision < 70; revision++) { - // Once a population removes an occurrence, it cannot overwrite that - // occurrence's head until it has rejoined the exact retained value. Keep - // the removed member absent while the remaining owned members move. - const original = await batch(revision > 35 ? 3 : 4, false, revision === 35 ? 3 : 4); - const request = parseContributionBatch({ ...original, mutations: original.mutations.map((mutation, index) => mutation.kind !== "put" ? mutation - : { ...mutation, row: { ...mutation.row, utcDay: DAY + (index + revision) % 2, client: (index + revision) % 2 ? "claude" : "codex", - tokens: { ...mutation.row.tokens, input: String(1000 + revision * 13 + index) } } }) }); - if (!request) throw new Error("invalid synthetic generated correction"); - await publish(request); - for (const mutation of request.mutations) { - if (mutation.kind === "put") expected.set(mutation.id, mutation.row); - else if (mutation.kind === "remove") expected.delete(mutation.id); - else throw new Error("unexpected synthetic mutation"); + // Keep this synthetic burst in one real Durable Object callback. Every + // revision still uses production reserve, journal, commit and apply paths; + // crossing the test RPC bridge for each small step only adds fixture I/O. + await runInDurableObject(stub(), async (_instance, context) => { + const localCanonical: typeof canonical = async run => run(new ContributionState(context.storage)); + for (let revision = 0; revision < 70; revision++) { + // Once a population removes an occurrence, it cannot overwrite that + // occurrence's head until it has rejoined the exact retained value. Keep + // the removed member absent while the remaining owned members move. + const original = await batch(revision > 35 ? 3 : 4, false, revision === 35 ? 3 : 4, DEVICE, localCanonical); + const request = parseContributionBatch({ ...original, mutations: original.mutations.map((mutation, index) => mutation.kind !== "put" ? mutation + : { ...mutation, row: { ...mutation.row, utcDay: DAY + (index + revision) % 2, client: (index + revision) % 2 ? "claude" : "codex", + tokens: { ...mutation.row.tokens, input: String(1000 + revision * 13 + index) } } }) }); + if (!request) throw new Error("invalid synthetic generated correction"); + await publish(request, localCanonical); + for (const mutation of request.mutations) { + if (mutation.kind === "put") expected.set(mutation.id, mutation.row); + else if (mutation.kind === "remove") expected.delete(mutation.id); + else throw new Error("unexpected synthetic mutation"); + } + const status = await applyAll(context.storage); + expect(status).toMatchObject({ appliedRevision: 4 + revision, publishedRevision: 3, appliedLag: 0, publishedLag: revision + 1 }); } - const status = await applyAll(); - expect(status).toMatchObject({ appliedRevision: 4 + revision, publishedRevision: 3, appliedLag: 0, publishedLag: revision + 1 }); - } + }); expect(now).toBe(started); expect((await snapshot()).publications).toHaveLength(2); expect(await cells(3)).toEqual(prior); const applied = await projected(state => state.control()); expect(applied.appliedRoot).not.toBeNull();