/* ============================================================ NYLA: case chapters, laid out natively in the portfolio's design language and registered into the existing tab shell. Nyla is the multi-self system that emerges from the evidence Nomi produced; the copy below keeps validated results, V1 hypotheses, and long-term vision clearly separated. The Prototype tab embeds the V1 product from assets/nyla-proto/ (a snapshot of the standalone build in ~/Desktop/Nyla-Real; re-copy those five files to update it). Registers into window.CASE_PANELS["nyla"]. ============================================================ */ (function () { const K = window.CSKit; const SectionHead = K.SectionHead; /* ---------- shared bits (same vocabulary as the nomi chapters) ---------- */ function H3Block(props) { return (

{props.title}

{props.sub ?

{props.sub}

: null}
); } function Badge(props) { return {props.children}; } function Chain(props) { return (
{props.nodes.map(function (n, i) { return ( {i > 0 ? → : null}
{n.t}{n.s ? {n.s} : null}
 ); })}
); } /* one metric row; pass from+to for a delta (tick marks the "from") */ function Metric(props) { const to = parseInt(props.to, 10); const from = props.from != null ? parseInt(props.from, 10) : null; return (
{props.label} {from != null ? props.from + " → " + props.to : props.to}
); } function MetricCard(props) { return (
{props.kicker}
{props.badge ? props.badge : null}
{props.p ?

{props.p}

: null}
{props.children}
); } function Down() { return ; } /* ============================================================ OPENING: positioning, the social-self, Nomi → Nyla ============================================================ */ function MetaCard(props) { return (
{props.label}
{props.value}
); } function OpeningPanel(props) { const c = props.case; return (

NYLA

Nyla is an AI-native system for managing the different versions of yourself you create and communicate through. It gives personal AI a missing identity layer: multiple contextual selves, each with its own memory, expression, and purpose.

The opening claim

Personal AI should know more than who you are. It should know which version of you belongs here.

Nomi tested whether one reusable social-self could improve AI-assisted creation. The next question was structural: if people authentically show up differently across contexts, why should personal AI represent them through one universal profile? Nyla expands the social-self into a system. Users maintain multiple contextual selves, give each one relevant memory, switch intentionally between them, and create through the combination of identity, memory, and context.

Nomi validates the object. Nyla explores the system.

AI should understand and preserve the different versions of a person instead of flattening them into one profile.

The ZeyaAI thesis. Nomi is the first product that tests it; Nyla is the system that follows the evidence.

{c.tags.map(function (t) { return {t}; })}
What is a social-self?

A chosen, contextual version of a real person

Generative AI dramatically increases the speed of creation, but it tends to compress people into one generic AI voice. People do not actually show up as one fixed identity: the ways someone communicates as a designer, a founder, a friend, or a private person can all be authentic and intentionally different. A social-self captures one of those versions: how you want to show up in a particular context.

{["a contextual version of the real user", "intentionally expressed in a specific situation", "persistent enough for AI to remember", "editable and always user-controlled", "able to accumulate relevant context"].map(function (t) { return (
{t}
); })}

Registers a user can choose to express a self in, never personality types the system assigns:

{["thoughtful", "professional", "intimate", "playful", "culturally tuned", "authoritative", "exploratory"].map(function (t) { return {t}; })}
A social-self is not a fictional character, an avatar, or a fake persona, and never a fixed personality archetype. Every self is the same real person, expressed for a different context, in the user's own words.
What Nyla is not

The framing that keeps the product honest

Nyla is not a social-media tool, not a generic AI assistant, and not another profile-management system. It sits at the intersection of AI-native identity systems, personal AI, creator infrastructure, contextual memory, and human-AI interaction.

{["AI-native identity", "personal AI", "creator infrastructure", "contextual memory", "human-AI interaction"].map(function (t) { return {t}; })}
The identity layer is the product. Creation, memory, and context organize around it.
Each arrow is earned by evidence, not added by ambition.
Nomi
validated wedge

One social-self, proven in a real workflow

Nomi's production wedge is RedNote content creation. A creator shapes one reusable social-self, creates through it, keeps a consistent way of showing up across posts, builds content history around that self, and reuses it instead of re-prompting AI every time. The purpose was never AI copywriting; it was to prove that a persistent social-self can be a meaningful control layer between a person and generative AI.

Nyla
unvalidated system

Many selves, each with memory and purpose

Nyla lets people create and own multiple social-selves, understand what each one is for, switch between them intentionally, build memory around each one, organize what each created, reuse previous context instead of rebuilding identity through prompts, and adapt one self across platforms while preserving continuity. Over time it can grow from a creator system into a broader contextual identity layer, but V1 stays anchored in identity, creation, memory, and context.

); } /* ============================================================ RESEARCH: the Nomi evidence, honestly read ============================================================ */ function ResearchPanel() { return (
validated} p="Did generating through a social-self beat direct generation at all?"> validated} p="The unprompted signal: people wanted the self to live beyond the session.">
What Round 1 revealed

The concept had pull, and the pull was bigger than style

The experience was still too broad across platforms, and the creation metrics hovered near a coin flip. But users responded far more strongly to save,{" "} reuse, and{" "} create another version than to any stylistic adjustment. That was the first evidence the opportunity was larger than one customized AI voice.

validated} p="Same comparison, one platform, qualified participants. Each tick marks Round 1; each bar is Round 2.">
Interpretation

Not every metric was a win, and that is the point

What strengthenedThe most fundamental social-self value improved sharply: intended-self alignment rose about 19 percentage points once the platform and audience got specific. Postability and editing reduction moved with it. A social-self improves the immediate creation experience when context is focused.
What did notSpeed stayed flat at 44%, and the persistence signals stayed strong but did not increase. Narrowing improved output value, not effort, so future value cannot come from more setup. It has to come from memory and reuse making contextual identity increasingly automatic.
The strategic question this producedIf one social-self is useful, what happens when users need different versions of themselves for different contexts? That question is where Nyla begins.
{[ ["Users preferred identity-shaped outputs", "AI needs a layer between generic user memory and output generation. That layer is the social-self."], ["Users wanted to save and reuse the self", "The social-self must be persistent: an object that survives sessions, not something reconstructed through prompting."], ["Users wanted another version of themselves", "Identity should be plural. One universal profile is the wrong data model for a person."], ["Platform specificity improved the core value", "A social-self must understand context, not merely tone: platform, audience, purpose."], ["Speed did not materially improve", "Future value cannot come from added setup. Nyla must make contextual identity automatic through memory and reuse."], ].map(function (row, i) { return (
Evidence {i + 1}

{row[0]}

Implication

{row[1]}

); })}
H1 through H5
unvalidated Nyla hypotheses
{[ ["H1", "Retained users will meaningfully use more than one social-self."], ["H2", "Different social-selves will accumulate meaningfully different memories."], ["H3", "Users will intentionally switch selves depending on context."], ["H4", "Accumulated memory will reduce repeated prompting and increase long-term utility."], ["H5", "The same social-self can adapt across platforms without collapsing into one generic cross-platform voice."], ].map(function (h) { return (
{h[0]}

{h[1]}

); })}
Nothing in this list is claimed as proven. The Testing tab holds the experiments each hypothesis needs, and the Outcome tab holds the behavior that would confirm them.
); } /* ============================================================ IDEATION: the problem, the object, the system model ============================================================ */ function IdeationPanel() { return (

Status marker for this whole tab: everything proposed below, the object model, the four layers, and the core mechanism, is unvalidated design intent for Nyla. Nothing here has been user-tested, so every badge in this tab marks a proposal rather than a finding.

Model A · the blank user

No memory, endless prompting

The model knows nothing about you, so every session starts from zero. You reconstruct who you are, what you sound like, and what the context is, every single time. Identity lives in the prompt, which means it evaporates when the session ends.

Model B · the single profile

One memory pretending to be a person

The model keeps one persistent profile representing "the user." But people are contextual: the designer, the founder, the friend, and the private person are all authentic and intentionally different. One memory collapses those differences into an average.

What both models cost the user

Blank context creates repetition. Universal memory creates context collapse. Nyla explores a third model: plural, contextual identity with explicit memory boundaries. That third model is the AI product thesis, and nothing in this tab is allowed to bury it beneath creator features.

{["repeated prompting", "generic output", "loss of distinctiveness", "context collapse", "inconsistent expression", "poor control over what AI remembers", "professional and personal contexts blur together"].map(function (t) { return {t}; })}
{[ ["Q1", "What belongs to each self?", "Memory, expression, and history need an owner, not a pile."], ["Q2", "What can be shared?", "Some things are the person, not the self. Sharing must be a choice."], ["Q3", "Which self is active?", "Switching has to be visible and intentional, never inferred silently."], ["Q4", "How does the system learn without silently redefining the user?", "Evolution must be proposed, reviewed, and accepted, never applied."], ].map(function (q) { return (
{q[0]}

{q[1]}

{q[2]}

); })}
social-self · the object
Nyla's proposed data model
{[["Purpose", "what this self is for"], ["Intended contexts", "the situations this self is declared for"], ["Expression", "voice, register, energy"], ["Preferences", "what it favors and avoids"], ["Boundaries", "what it will not say or share"], ["Memory", "what it has learned, inspectable"], ["Content history", "everything created through it"], ["Platforms", "where it shows up"], ["Evolution", "how it changes over time"]].map(function (f) { return (
{f[0]}
{f[1]}
); })}
This turns identity from a prompt into persistent infrastructure. Intended contexts is the scope the user declares for a self; it is not the runtime CONTEXT layer below, which is the single situation the active self is speaking into right now.
{[ ["PERSON", "one real person, always in control", false, []], ["SOCIAL-SELF LAYER", "the versions of you, created and named by you", true, ["Founder", "Personal Storyteller", "Creative Explorer", "Industry Voice"]], ["MEMORY LAYER", "what each self knows, remembers, and is allowed to use", true, ["self-specific", "shared intentionally", "general"]], ["CONTEXT", "where and to whom the active self is speaking right now", true, ["platform", "audience", "task"]], ["CREATION", "what each self creates in its own contexts", true, []], ].map(function (l, i) { return ( {i > 0 ? : null}
{l[0]} {l[1]}
{l[3].length ? (
{l[3].map(function (t) { return {t}; })}
) : null}
); })}
↻ learning over time: accepted edits and reuse refine each self's memory, never the person's identity

Nyla does not decide who you are. It suggests, remembers, and adapts; the person always chooses which self acts, and can override the suggestion at any time.

The core mechanism
Every Nyla output is this combination. Change any one input and the creation changes with it; that is what makes identity a control layer instead of a style preset.
I did not begin by building the largest vision. I isolated the riskiest assumption and let the evidence set the scope.
); } /* ============================================================ TESTING: the success model Nyla will be judged against ============================================================ */ function TestingPanel() { const LEVELS = [ { k: "Level 1 · Social-self value", badge: validation underway via Nomi, q: "Does creating through a social-self beat generic generation?", m: ["intended-self fit", "postability / usability", "editing reduction", "preference vs. direct generation"], note: "Nomi has already begun validating this layer: +19 pts intended-self fit in Round 2.", }, { k: "Level 2 · Persistence", badge: Nyla V1 hypothesis, q: "Does a social-self become something users return to?", m: ["social-self reuse rate", "artifacts created per saved self", "% of active users returning to the same self", "time between self creation and first reuse"], note: "Nomi's 75–81% stated save intent makes this the first behavior V1 must convert from stated to observed.", }, { k: "Level 3 · Multiplicity", badge: Nyla V1 hypothesis, q: "Do users genuinely need more than one version of themselves?", m: ["% of retained users creating a second self", "% actively using 2+ selves", "average active selves per retained user", "contexts differentiated between selves"], note: "This is Nyla's strongest early product-market signal: plurality is the premise the whole system rests on.", }, { k: "Level 4 · Contextual network value", badge: future, q: "Does Nyla get more useful as a user's history grows?", m: ["30- and 90-day retention", "creation frequency", "cross-platform reuse", "% of new content drawing on existing memory", "reduction in repeated prompting"], note: "This is where Nyla stops being a feature and starts being infrastructure.", }, ]; return (
{LEVELS.map(function (l, i) { return ( {i > 0 ? : null}
{l.k}
{l.badge}

{l.q}

{l.m.map(function (m) { return {m}; })}
{l.note}
 ); })}
{[ ["Test 1 · Multiplicity", "Can users meaningfully distinguish two social-selves?", ["second-self creation", "actual use of the second self", "switching behavior"], "tests H1 + H3"], ["Test 2 · Memory separation", "Does self-specific memory improve relevance without creating confusion?", ["memory acceptance", "memory removal", "context mistakes", "user trust"], "tests H2 + H4"], ["Test 3 · Context switching", "Does selecting a different self produce meaningfully preferred outputs for different contexts?", ["intended-self fit", "editing burden", "preference vs. one generic voice", "reuse"], "tests H3 + H5"], ].map(function (t) { return (
{t[0]}
not yet run

{t[1]}

{t[2].map(function (m) { return {m}; })}
{t[3]}
); })}
Monthly Multi-Self Creators

Retained users repeatedly create through multiple saved social-selves

A retained user who uses at least two distinct saved social-selves to complete meaningful creation flows during a month. One behavior that jointly requires retention, persistence, multiplicity, and real social-self value, none of which can be gamed by the others.

Deliberately rejected as north stars: account creation, number of selves created, raw generation volume. All three can grow while the product fails.
); } /* ============================================================ ITERATIONS: Nyla V1, deliberately narrow ============================================================ */ function IterationsPanel() { const SELVES = [ ["Industry Voice", "professional analysis, measured"], ["Personal Storyteller", "warm, first-person, unhurried"], ["Creative Explorer", "playful, visual, experimental"], ["Close-Friends Self", "unpolished, private, direct"], ["Founder Self", "building in public, candid"], ]; return (

Status marker for this whole tab: V1 is a proposed scope, not a shipped one. Every "V1 core" and "V1 principle" badge below marks an unvalidated design decision waiting on the experiments in the Testing tab.

A · Multi-self management
V1 core

Create, name, switch, evolve, and always see how selves differ

Users create a new social-self, save it, name it, describe its purpose, and see where it is used. They can switch between selves intentionally, duplicate or evolve an existing one, and compare how each differs. Names are the user's own words, never fixed personality archetypes. The five below are example names a user wrote, not a set Nyla ships:

{SELVES.map(function (s) { return ( {s[0]} {s[1]} ); })}
B · Self-specific memory
V1 core

Each self remembers its own world, and shows its work

Memory belongs to a self, not to a global profile. Each self accumulates:

{["past content", "recurring topics", "preferred language", "audience", "examples the user liked", "content patterns", "important experiences", "vocabulary preferences", "boundaries"].map(function (t) { return {t}; })}
Memory is never invisibleUsers can always see what Nyla remembers, which self remembers it, where it came from, and why it is being applied to the current creation, and they can edit or remove any of it. This is the line between contextual memory and an opaque profile.
{[ ["a private reflection", "Personal Storyteller only"], ["a professional project detail", "Industry Voice only"], ["a general language preference", "shared across selves, by choice"], ].map(function (r) { return (
{r[0]} → {r[1]}
); })}

The user's controls over any memory:

{["keep", "edit", "delete", "share", "move", "separate"].map(function (t) { return {t}; })}
The governing principle: AI should remember context without collapsing contexts. Not everything the AI knows about the user belongs to every self, and nothing here is presented as invisible magic.
C · Contextual creation
V1 core

One idea, three selves, three honest versions

Creation starts from an idea, and Nyla helps answer three questions: which social-self should carry it, what context from that self matters, and how the idea should change for this audience and platform. The same idea, say "what I learned shipping a product in six weeks", becomes:

{[ ["Founder Self", "LinkedIn", "a professional post: the decision framework, the trade-offs, what the team would repeat"], ["Personal Storyteller", "RedNote", "a thoughtful note: the 2am doubts, the small rituals that kept the weeks moving"], ["Creative Explorer", "Instagram", "a concise caption: one image, one sharp line about making things fast"], ].map(function (r) { return (
{r[0]} → {r[1]}

{r[2]}

); })}
Three versions, one person: in each, a different self and its own memories are in play, so the idea adapts without flattening into a generic cross-platform voice. This context switch is the signature Nyla interaction, and the Prototype tab makes it operable.
{["profile-picture generation", "avatar generation", "generic image generation", "LinkedIn cold messages", "calendar management", "email assistant", "generic life assistant", "task manager", "full autonomous agent"].map(function (t) { return {t}; })}

Nomi's original ideation explored networking messages and profile appearance. Those were useful exploratory branches, but a branch does not return as a Nyla feature simply because it existed earlier; it returns only if the identity system proves out and the branch serves it.

Why not profile pictures yet

A different product problem, not a smaller one

Profile-image generation introduces visual identity, image-model quality, likeness, safety, and aesthetic consistency: an entire problem space of its own. None of it validates contextual social-self management, so it stays outside the first Nyla proof.

Why not LinkedIn messaging yet

The coherent expansion is a self, not a message type

LinkedIn is already a V1 publishing context: the Founder Self posts there, and the prototype switches the same idea between LinkedIn and RedNote. What V1 defers is outbound messaging and networking. "LinkedIn messages" is a message type rather than a self, so it is too narrow to build on. That work returns as a Professional social-self carried into professional creation and communication contexts, and only after multi-self behavior is validated.

{[ ["One social-self", "Multiple social-selves", "the one expansion V1 makes, because plurality is the thesis itself"], ["Universal memory", "Scoped, per-self memory", "context collapse is the failure mode Nyla exists to prevent"], ["Platform as product architecture", "Platform as context", "a platform is a room the self speaks in, not a product line"], ["Manual configuration", "Learning through accepted edits", "the flat speed metric proved setup cannot be the price of identity"], ["More features", "A stronger identity system", "every feature rejected above buys conceptual clarity for the proof"], ].map(function (row) { return (

{row[0]}

{row[1]}

{row[2]}

); })}
{[ ["You create and name your selves", "Nyla never generates a persona and assigns it to you."], ["Memory is inspectable", "Every self shows what it remembers; you can change or delete any memory."], ["Selves never merge silently", "Sensitive contexts stay separated unless you explicitly connect them."], ["Suggestions, not impositions", "Nyla may suggest which self fits a moment; it never imposes one."], ["Override at any time", "The selected self is always yours to change, mid-flow included."], ["Boundaries are respected", "What a self will not say or share is part of the self, and it holds."], ].map(function (p) { return (

{p[0]}

{p[1]}

); })}
Learning through use
V1 principle

The long-term value is not additional setup. It is decreasing setup.

Example: the user keeps removing promotional wording from Industry Voice outputs. Nyla notices the pattern across three posts and asks, instead of acting:

You made the last three Industry Voice posts more understated. Update this social-self?

{["Yes", "Not now", "Review"].map(function (t, i) { return {t}; })}
Social-selves evolve through behavior, not endless onboarding forms, and never silently. This is the same answer as Q4 in Ideation, expressed as an interaction.
); } /* ============================================================ OUTCOME: expansion, business model, defensibility, ledger ============================================================ */ function OutcomePanel() { return (
Expansion model
Each stage is unlocked by the success model's levels: Stage 2 requires Level 3 multiplicity to be real. The path never jumps from a creator workflow to "manage your entire life": the long-term vision is credible only after contextual identity proves valuable beyond content.
Supporting metrics
{["social-self reuse", "second-self adoption", "% of users actively using 2+ selves", "context-switch frequency", "memory reuse in new creations", "30-day retention", "90-day retention"].map(function (t) { return {t}; })}
Why these metrics decide the product
{[ ["Users only ever create one self", "Nyla may not need to exist separately from Nomi."], ["Many selves created, none reused", "The product is novelty, not a system."], ["Repeated creation through different saved selves", "Multiplicity has real value."], ["Memory makes those selves increasingly useful", "Nyla begins accumulating product value."], ["Users return because the system remembers contextual identity", "Nyla becomes hard to replace with a generic prompt."], ].map(function (r) { return (
{r[0]} → {r[1]}
); })}
Free

One primary social-self

Limited memory, limited creation. This tier already exists in the world: it is Nomi, and it doubles as the top of the funnel and the ongoing Level-1 validation surface.

Nyla Individual

The multi-self system

Multiple social-selves, persistent contextual memory, cross-platform adaptation, deeper history, and advanced creation workflows. The subscription maps exactly to the behaviors the north star measures.

Future · Professional / Creator

When selves become work

Analytics, team and collaborator workflows, brand and professional contexts, deeper knowledge integration. Deliberately underdeveloped until the individual tier proves out.

Every interaction is designed to emit structured signals
{["which self", "which context", "which memory was used", "what the user selected", "what the user edited", "accepted vs. rejected outputs", "reuse", "platform adaptation", "switching between selves"].map(function (t) { return {t}; })}
A longitudinal contextual identity graph

Nyla is designed so that every use would deepen its read of which self is appropriate, which memory matters, how the user expresses each self, and where the boundaries between selves must hold. Corrections and preference signals would compound per self, producing value a single flat prompt history structurally cannot capture. No user has built this graph yet; it is the advantage V1 is meant to earn, not one the product currently holds.

Validated
Nomi, n=32

A social-self measurably improves intended-self fit, postability, and editing burden when context is focused. Users state strong intent to save and to create more selves.

Hypothesis
Nyla V1 will test

That stated persistence converts to reuse behavior; that retained users genuinely need 2+ selves; that memory makes contextual identity automatic instead of adding setup.

Future vision
unproven, by design

Cross-platform adaptation, contextual network value, and a personal AI operating layer. Presented as the expansion path the evidence points toward, not as achieved outcomes.

{[ ["What is a social-self?", "A chosen, contextual version of a real person: how you want to show up in one context. Not a character, avatar, or fake persona."], ["Why is one AI profile not enough?", "People are contextual. One memory averages the designer, the founder, and the friend into a voice that belongs to none of them."], ["What did Nomi prove?", "Creating through a social-self beats direct generation on the metrics that matter: intended-self fit rose ~19 points when context got specific, and users want to save and multiply their selves."], ["What did Nomi not prove?", "That stated persistence becomes real reuse, and that speed improves. Both are Nyla V1's burden, which is why V1 leads with memory and reuse rather than more setup."], ["What does Nyla do first?", "Three things only: multiple named selves, visible per-self memory, and contextual creation that adapts one idea across audiences without flattening it."], ["How is it different from AI memory or a writing assistant?", "Memory is plural, per-self, inspectable, and user-controlled; identity is the product object, not a hidden profile behind an assistant."], ["What behavior proves it works?", "Monthly Multi-Self Creators: retained users creating through at least two distinct saved selves every month."], ["Why is this a platform, not a feature?", "Because it is built so that every use would deepen a structured, per-self identity graph, and each proven level would unlock a wider context where the same selves apply. That compounding is the advantage V1 is meant to earn, not one it has yet."], ].map(function (q) { return (

{q[0]}

{q[1]}

); })}
); } /* ============================================================ PROTOTYPE: the V1 product, running locally in the page Nyla is a full operating environment (persistent sidebar, internal scrolling), so it is framed at a fixed viewport height and scrolls inside, rather than auto-growing the way the single-screen Nomi flow does. ============================================================ */ const PROTO_SRC = "assets/nyla-proto/index.html"; /* Each jump lands on a screen the flow below starts from. */ const PROTO_JUMPS = [ ["Home", "home"], ["Selves", "selves"], ["A social-self", "self/iv"], ["Memory", "memory/ps"], ["Create", "create"], ]; /* The site's own light/dark toggle drives the prototype. The panel doesn't receive the theme as a prop, so it watches the attribute the shell sets. */ function useSiteTheme() { const read = function () { return document.documentElement.getAttribute("data-theme") === "dark" ? "dark" : "light"; }; const [theme, setTheme] = React.useState(read); React.useEffect(function () { const obs = new MutationObserver(function () { setTheme(read()); }); obs.observe(document.documentElement, { attributes: true, attributeFilter: ["data-theme"] }); setTheme(read()); return function () { obs.disconnect(); }; }, []); return theme; } function PrototypePanel() { const frameRef = React.useRef(null); const [jump, setJump] = React.useState("home"); const theme = useSiteTheme(); /* Baked into the initial src so the prototype boots in the right theme with no flash. Captured once: changing src later would reload and lose state. */ const [bootSrc] = React.useState(function () { return PROTO_SRC + "?theme=" + (document.documentElement.getAttribute("data-theme") === "dark" ? "dark" : "light"); }); /* Every later toggle is pushed in: the direct call when same-origin lets us skip a frame, postMessage covers file:// and any cross-origin host. */ React.useEffect(function () { const ifr = frameRef.current; if (!ifr) return; ifr.style.colorScheme = theme; try { const win = ifr.contentWindow; if (win && typeof win.nylaSetTheme === "function") { win.nylaSetTheme(theme); return; } if (win) win.postMessage({ type: "nyla-theme", theme: theme }, "*"); } catch (e) { try { ifr.contentWindow.postMessage({ type: "nyla-theme", theme: theme }, "*"); } catch (e2) {} } }, [theme]); function go(hash) { setJump(hash); const ifr = frameRef.current; if (!ifr) return; /* Same-origin: navigate in place so nothing reloads. If the page is opened straight off the filesystem the access throws, so fall back to src. */ try { const win = ifr.contentWindow; if (win && win.location) { win.location.hash = "#" + hash; return; } } catch (e) {} ifr.src = PROTO_SRC + "?theme=" + theme + "#" + hash; } return (
Jump to {PROTO_JUMPS.map(function (j) { const on = jump === j[1]; return ( ); })}