CueCrux Platform

Ask. See the proof. Know with certainty. That’s CueCrux.

Answers you can prove. Receipts, not vibes.

We built it so that when you ask a question online, you don’t just get an answer; you get the evidence that makes it true.

CueCrux is more than a search tool; it’s a new way to trust information.

Every response arrives with citations, timestamps, and provenance details, revealing why an answer exists, not just what it says.
It transforms knowledge from something you take on faith into something you can verify, inspect, and keep.

At a glance

• Proof outputs by default: citations, quote spans, timestamps, licences, provenance.
• QUORUM (MiSES): minimal evidence sets per claim; no citation spam.
• Modes that match the stakes: light, verified, audit (with visible rigour and cost).
• Replay and counterfactuals: re-run later, diff what changed, test “what if this source is removed?”.
• Trust signals you can scan: domains, freshness, coverage, contradictions, Trust Score.
• Independent oversight: WatchCrux audits the stack and publishes PASS/WARN/FAIL findings.

A New Kind of Knowledge Platform

CueCrux isn’t just a search interface or another AI assistant; it’s the visible layer of a new kind of information ecosystem that values quality of data over volume of data.

Where traditional web crawlers collect everything and hope relevance emerges later, CueCrux collects less, but proves more.
Every artefact it stores is licence-aware, timestamped, and verified at the moment of capture, creating a foundation built not on scale, but on integrity.

Beneath the surface runs the CueCrux Protocol: an open, evidence-first data protocol that treats every interaction as a traceable event in a distributed ledger of knowledge.

Each entry is signed, versioned, and linked to its source, ensuring that information cannot be altered without a visible trail.
Instead of indexing the web by popularity or advertising reach, CueCrux curates a living graph of verifiable artefacts: content that has passed policy, provenance, and authenticity checks.

Where conventional systems return cached snapshots or opaque model outputs, CueCrux enforces a higher standard:
every claim must be anchored to verifiable evidence, every answer reproducible, and every change accountable.
It’s not about finding more data; it’s about finding the right data, proving it once, and trusting it always.

This approach redefines how data is stored and delivered:

• Storage with proof: Every artefact is hashed, signed, and versioned through the CROWN trust framework.
• Compute with accountability: Queries run within cost-conditioned envelopes (C³) that produce deterministic receipts rather than transient model outputs.
• Delivery with context: Users receive structured answers that can be replayed, verified, and cited across time, forming a living, evolving record of truth.

Vision and Purpose

The broader vision of CueCrux is to create a world where digital knowledge is provable, portable, and permanent.
Instead of trusting an algorithm’s reputation, you trust the evidence behind its answers.

Instead of scattered datasets and unverified sources, you operate within a transparent ecosystem where every answer from a quick fact to an audited report carries a cryptographic proof of origin.

The upcoming CueCrux Protocol Whitepaper will formalise this model: defining how artefacts, receipts, and provenance chains interact across Engine, Factory, Watch, and Web planes to make knowledge storage verifiable by design and tamper-evident by default.

In short, CueCrux is not only a platform for answers; it’s a step toward a more accountable web, where credibility scales with proof, and every user, creator, or organisation can finally know with certainty.

Evidence First, Not Vibes

Each answer in CueCrux is built from a QUORUM (MiSES): a MiSES (Minimal Evidence Set) assembled by the Quorum of Unified Observations and Referenced Underlying Material (QUORUM) process: the smallest, non-redundant collection of independent citations required to support the full chain of reasoning behind a claim.
This isn’t a stylistic choice; it’s a foundational rule of the CueCrux Protocol. When a user asks a question, the system searches not only for relevant information but for the minimal set of independent sources that can jointly substantiate each statement.

Behind the scenes, QUORUM (MiSES) acts like a mathematical proof assistant for language:

• It decomposes the answer into atomic claims, ensuring each one has traceable support.
• It scores potential citations based on independence, freshness, and reputation, selecting only those that strengthen the overall certainty of the claim.
• It avoids duplication two sources repeating the same statement don’t double the trust score.
• It seeks counter-evidence automatically, highlighting where reliable disagreement exists rather than filtering it out.

The result is an answer that behaves more like a scientific paper than a chat response every claim, timestamp, and domain is verifiable, and contradictions are surfaced transparently.

QUORUM turns opaque generative output into an auditable knowledge object, ready for citation, replay, or inclusion in future workflows.

Who CueCrux Is For

CueCrux is built for anyone who believes decisions should be based on proof, not probability.

Whether you’re investigating, building, learning, or communicating, the platform gives you the same ability to know with certainty scaled to your needs and context.

For the Public - Anyone Seeking Clarity

At its core, CueCrux is for everyone. You don’t need to be an academic, engineer, or policy maker to want better answers. Anyone can ask a question, verify the source, and understand where information truly comes from. It’s a tool for collective certainty, a step toward a web where transparency is the default and trust is measurable.

For Analysts, Researchers, and Journalists

Professionals who live and die by the accuracy of their information. CueCrux gives them an environment where every answer carries citations, timestamps, and replayable evidence. Instead of manually checking references or worrying about source bias, they can inspect the provenance of each statement directly in the UI, saving hours of verification time while increasing editorial confidence.

For Teams and Organisations

Departments that rely on shared understanding (strategy, compliance, or product development) can use CueCrux as a single, verifiable knowledge surface. Workspaces let teams save, tag, and re-run answers with full context and mode history, ensuring decisions are based on evidence, not outdated slides. For regulated or safety-critical industries, audit mode provides permanent cryptographic receipts that satisfy internal and external review requirements.

For Developers and Data Owners

Technical users who manage APIs, datasets, or custom integrations can use CueCrux to ingest and verify their own sources through FactoryCrux. Licences, robots.txt, and provenance are honoured automatically, preventing accidental misuse of restricted data. SDKCrux clients allow programmatic access to verified answers, receipts, and audit metadata, enabling developers to build trusted experiences into their own products.

For Support and Operations Teams

SupportCrux extends CueCrux into customer service: every automated or human-generated reply carries proof of its claims. This eliminates “black box” responses and provides managers with verifiable metrics such as response confidence and citation coverage. Operations staff benefit from WatchCrux oversight and OpsCrux dashboards, giving full visibility into accuracy, uptime, and cost performance.

For Educators, Students, and Learners

CueCrux is also a learning environment where curiosity meets verification. Students can explore complex topics and instantly see which statements are supported, disputed, or unverified. Educators can design assignments that emphasise critical thinking, teaching how to reason with evidence, not just recall information.

For Builders, Innovators, and Entrepreneurs

Start-ups and independent creators can use CueCrux to ground their ideas in solid research. By referencing verified answers, they can reduce uncertainty in product development, marketing, or investment materials. Every citation becomes a trust anchor that improves communication and credibility.

In short: CueCrux is for anyone who wants to think, write, build, or decide with confidence.
The goal isn’t just accurate information, it’s a culture of verifiable understanding.

Core Concepts Explained

Before diving into how WebCrux feels to use, it helps to understand the three pillars that make it work:
CROWN, QUORUM (MiSES), and Receipts, all of which revolve around a single building block: the Artefact (stored in the Perofant™ evidence database).

Artefacts - The Building Blocks of Knowledge

An Artefact is the smallest verifiable unit of knowledge inside the CueCrux ecosystem.
Think of it as a preserved digital fingerprint of a piece of evidence - a paragraph from a webpage, a figure in a report, a policy line from a PDF, or even a dataset record.

Each artefact is more than just the text or data it contains. It carries a complete metadata envelope that makes it auditable, comparable, and trustworthy.

What an Artefact Holds

1. Main Data Payload
   The literal piece of content that was ingested for example, a paragraph, table, or snippet of code.
   This payload is stored in a normalised form (plain text or structured JSON) to ensure it can be processed consistently across regions and updates.
2. Metadata Fields
   Every artefact includes:
   • Source URL or document ID – where it came from.
   • Observed timestamp – when it was first captured and last verified.
   • Content hash – a BLAKE3 or SHA-256 fingerprint that proves immutability.
   • Author and publication details – when available.
   • Licence and jurisdiction – defining how it can be reused.
   • Reputation score – the trustworthiness of the source domain.
   • Version number – incremented whenever the artefact is re-ingested or superseded.
   • Parent/child links – connecting related artefacts into provenance trees.
3. Provenance Ledger Entry
   Each artefact is cryptographically signed and recorded in the Provenance Ledger a tamper-evident chain that forms the foundation of CueCrux’s transparency. This ledger allows anyone to prove that a citation or quote truly existed at a particular moment in time, even if the original website changes or disappears.

Artefacts are append-only: they cannot be changed, only versioned.
When an update occurs, a new artefact is created and linked to the old one, forming a historical chain of custody.
Over time, these chains merge into what CueCrux calls the Global Provenance Graph a living record of how knowledge evolves.

Perofant™ - The Evidence Database (Proof-First Artefact Network)

CueCrux stores artefacts in the Perofant™ evidence database: a proof-native data plane built for provenance-first retrieval at scale.
Perofant is where artefacts live (hot/warm/cold tiers), while the Provenance Ledger records the signed events that make those artefacts auditable (capture, verification, receipts, policy decisions, and updates).

Key properties:

• Evidence is first-class: artefacts keep licence, robots, jurisdiction, and integrity metadata alongside the payload.
• Hot / warm / cold tiers: frequently used artefacts stay fast; long-tail material can move without breaking receipts.
• Version lineage, not silent rewrites: updates create new artefacts linked to prior versions, preserving chain-of-custody across time.

CROWN - The Trust Backbone

CROWN stands for Counterfactual, Reputation-Weighted, Origin-Verified, Workflow-Encoded four interlocking principles that govern every answer CueCrux produces.

1. Counterfactual

CROWN doesn’t just look for confirming evidence; it also searches for credible contradictions.
By deliberately seeking out opposing viewpoints from reputable sources, CROWN ensures that uncertainty isn’t hidden it’s quantified and shown to the user.
In an Audit run, both supporting and counter-evidence appear side by side, giving a 360° view of the claim.

2. Reputation-Weighted

Each domain, author, or dataset carries a reputation prior, based on its historical accuracy, recency, and licence integrity.
When multiple artefacts support a claim, their weights are aggregated so that reliable, up-to-date, and well-licensed sources count more than unverified or outdated ones.
This helps prevent low-quality content or coordinated misinformation from dominating results.

3. Origin-Verified

Every artefact used in a CROWN assembly is signed at capture time using ed25519 cryptography and stored in the Provenance Ledger.
This guarantees that the system or any independent auditor can later prove that the cited content truly existed in that form and at that time.
No source can be forged, replaced, or silently edited without detection.

4. Workflow-Encoded

Finally, CROWN records the entire decision-making process that led to an answer.
This includes:

• The retrieval model and parameters used.
• The selected artefacts and their trust scores.
• Any counterfactual searches performed.
• The model version and cost envelope.

All of this is packaged into a CROWN Snapshot, a reproducible workflow that can be re-run in the future to confirm or challenge the result.

Together, these four layers ensure that every answer is grounded in verifiable proof, not transient computation.

QUORUM (MiSES) - Quorum of Unified Observations and Referenced Underlying Material

While CROWN defines how trust is built, QUORUM (MiSES) defines how much evidence is enough.
The concept comes from logic and mathematics: find the smallest set of inputs required to prove a statement beyond redundancy.

In CueCrux terminology, the output of QUORUM is a MiSES (Minimal Evidence Set): the minimal, non-redundant evidence bundle that supports the answer’s claims.

When WebCrux assembles an answer:

1. The system breaks the answer into atomic claims small, precise statements.
2. It searches for all artefacts that can support each claim.
3. It measures independence (are the sources genuinely different?), coverage (does the claim have full support?), and marginal utility (does adding this source add new information?).
4. It selects the minimal set that satisfies all claims: the MiSES (produced by QUORUM).

This keeps answers concise but also transparent: users can click to see exactly which sources were chosen and why others were excluded.
If the system detects disagreement, the counterfactual artefacts are shown in parallel, clearly marked as contradictory evidence rather than suppressed.

QUORUM effectively connects artefacts into small, logical clusters micrographs of agreement and disagreement that form the building blocks of verifiable answers.

Receipts - Verifiable Proof

Every time an answer is generated especially in Audit mode WebCrux creates a Receipt.
This is a digitally signed document that encapsulates:

• The question asked and trust mode used.
• The CROWN snapshot (model version, retrieval parameters, and weights).
• The list of artefacts (their hashes, timestamps, and licences).
• The MiSES groupings (which sources supported which claims, and how the QUORUM selection was formed).
• The counterfactuals (if any opposing evidence was found).
• The signatures and cost envelope that prove authenticity and accountability.

Receipts are stored immutably in the ledger and can be replayed at any time to reproduce the same answer or verify that its evidence still holds.
For enterprises, they serve as cryptographic audit trails; for individuals, they provide confidence that what they’re reading is grounded in reality.

How It All Connects

When you ask a question in WebCrux:

1. The Engine retrieves candidate artefacts from Perofant (the evidence database) and the provenance ledger.
2. CROWN evaluates them for reliability, counterfactual diversity, and authenticity.
3. QUORUM (MiSES) filters this pool down to the smallest non-redundant evidence set that covers every claim.
4. The resulting evidence, model workflow, and trust weights are sealed into a Receipt.

The process turns ordinary search into a transparent proof chain one where you can always trace back why an answer exists, where it came from, and when it was verified.

In essence:
Artefacts capture the facts.
Perofant stores and versions them.
CROWN governs their trust.
QUORUM (MiSES) keeps evidence minimal.
Receipts make it all provable.

Transparent Trust Modes - Why Three Levels Matter

CueCrux operates in a world where not every decision needs the same level of scrutiny. Sometimes you need speed, sometimes you need proof. CueCrux therefore provides three graduated modes of assurance so users can match response rigour to context and cost.

• Light - Ideal for exploration and brainstorming.
  In this mode, the system prioritises responsiveness, returning high-confidence results without performing full provenance or signature verification. It’s fast and inexpensive, perfect for quick iteration a “draft of trust” that can later be escalated.
• Verified - The platform’s default mode and the recommended baseline for publishing or operational use.
  Every citation in the Quorum set (MiSES) is checked for source authenticity, licence validity, and timestamp alignment. Independent domains are required wherever available. This strikes the balance between speed and reliability fast enough for daily use, but rigorous enough to stand behind.
• Audit - The gold-standard mode for compliance, safety-critical work, or public accountability.
  In audit mode, CueCrux activates the full Counterfactual Lane, deliberately searching for credible contradictions and running deterministic verification against the provenance ledger.
  A cryptographic receipt - signed via Vault Transit under the CROWN framework captures the entire evidence chain, query parameters, model version, and cost envelope, enabling replay at any time.

These modes exist because truth has contexts. A quick answer during research shouldn’t consume the same computational and verification budget as a regulated audit, yet both should live on the same transparent spectrum of trust.

By making rigour a controllable setting rather than a hidden process, CueCrux gives users informed choice not blind faith in the knowledge they rely on.

These are powered by C³ - Cost-Conditioned Cascades, ensuring reliable performance within predictable budgets and routing queries across light, verified, and audit based on explicit budgets and risk.

Separation of Powers - Trust by Architecture

CueCrux deliberately splits responsibilities across planes so no single component can quietly bend the truth:

• EngineCrux: holds proofs, assembles Quorum sets (MiSES), and issues receipts.
• WebCrux: the UX + auth layer; browsers never see Engine keys (BFF by default).
• WatchCrux: independent audits and drift detection, reported as PASS/WARN/FAIL.
• OpsCrux: budgets, cost-conditioned cascades, and operational dashboards.

This is a governance constraint as much as a technical pattern: proofs are append-only, and corrections/takedowns are versioned forward rather than silently rewritten.

The User Experience

Sign in → Ask → Skim answer → Inspect (Why Trust) → Save/Share → History/Workspace → Replay.

Every journey through CueCrux follows this same flow: simple on the surface, but deeply traceable underneath.

Each step is designed to prioritise clarity, proof, and permanence, ensuring that answers don’t just exist in the moment, but become part of a living, auditable record of knowledge.

1. Ask

Users pose a question in plain language, selecting their desired trust mode: light, verified, or audit. Under the hood, the CueCrux Engine retrieves evidence from its distributed artefact ledger, respecting regional data residency, licence boundaries, and visibility rules.

Whether the query is public or runs inside a private tenant plane (via the Private Stack), the system applies the same CROWN logic: provenance first, opinions later.

For enterprise or regulated deployments, queries can run entirely within a siloed environment, keeping artefacts and receipts confined to a local jurisdiction while still participating in the global proof network.

Each query produces not just an answer, but an evidence graph a cryptographic trail of what was referenced, when it was observed, and under which legal and geographical context it was stored.

2. Review

The user receives a concise, cited answer composed from one or more QUORUM (MiSES) evidence sets. Results are formatted for readability but backed by structured provenance data.

In Verified and Audit modes, a small “Proof Capsule” icon links directly to the CROWN record in the provenance ledger the authoritative source of what the system actually saw at the time of answering.

3. Inspect

By expanding the “Why Trust” panel, users can walk the reference tree of each artefact.
They can see:

• Which domains, documents, or datasets contributed to the claim.
• The timestamp and licence of every source.
• Any counter-evidence or contradictory findings.
• The lineage of each artefact when it was first indexed, if it was refreshed, and what version of the content was used.

Artefacts cannot be modified retroactively; they can only be versioned forward.

This immutability guarantees that if a source changes or disappears, the original record remains provable and auditable.
Users can also pivot from a single answer to explore related artefacts, compare them across regions or time periods, and see how factual context evolves.

4. Save & Share

Every answer regardless of mode can be saved to a personal library or shared into a workspace.

Workspaces support role-based collaboration: individuals can annotate, discuss, or link answers to projects while retaining the original provenance intact. Saved answers store both their content and their receipt identifiers, so that any collaborator can independently verify authenticity or regenerate an updated version in the future.

For organisations using Private Stack planes, shared answers can remain completely internal while still interoperating with the global CROWN registry.
This creates a federated trust model data stays local, but proof scales globally.

5. History & Replay

The History tab in WebCrux is more than a simple activity log it is a versioned timeline of certainty.
Every question you ask, and every answer the platform provides, becomes part of your personal and organisational provenance record.
Instead of static results, you build a living archive of how information and the evidence behind it has evolved over time.

When you revisit the History view, each saved answer can be explored or replayed.
Replaying means asking the same question again, but under today’s conditions: with the latest data, sources, and trust checks.
This allows you to see not just what the answer is now, but how it has changed since the last time you asked.

Re-run Past Queries in Any Trust Mode

WebCrux lets you choose how deeply to verify your answers, through three trust modes:

• Light mode replays quickly, ideal for checking if anything obvious has changed.
• Verified mode repeats the evidence verification process, confirming sources and timestamps.
• Audit mode goes further, re-examining all available counter-evidence and producing a new signed receipt that records today’s proof.

By switching modes, users can decide how much rigour to apply without losing their original record. Each run is stored as a new entry in the timeline, ensuring nothing is overwritten or hidden.

Filter by Date Range

Sometimes, truth depends on timing.
WebCrux allows users to look at answers as they existed within specific dates or time periods.
For instance, you might compare the guidance on a policy “as of March 2024” versus “as of today” to see exactly when the official stance changed.
The system stores both versions, giving you a clear view of how facts - and their evidence have shifted.

Filter by Source Region

Different parts of the world often hold different perspectives, data access laws, or official records. The source region filter lets you limit answers to evidence gathered from specific countries or jurisdictions.

For example, an environmental scientist could view citations only from European sources for compliance work, or include global sources to see a broader context.

This ensures every user, from local journalist to global researcher, can tailor provenance to their geographic or legal needs.

Filter by Licence Type

Not all data is created equal or equally shareable. WebCrux tracks the licence attached to every artefact, whether it’s public domain, Creative Commons, or restricted-use.
Filtering by licence type helps users understand what they can safely quote, publish, or reuse.

For example, you might filter to “open data” when building a report for redistribution, or “internal-only” when dealing with proprietary research.
This keeps intellectual property respected and re-use legally defensible.

Compare Evidence Over Time

When an answer is replayed, WebCrux automatically compares the evidence graphs the network of sources and citations behind each version.
It highlights:

• Which documents or websites have been added or removed.
• When certain facts were updated or retracted.
• How the balance of independent sources has changed.

This comparison provides a visible evolution of truth showing how knowledge itself adapts, matures, or corrects course over time.

Versioning and Chain of Custody

Every replay creates a new version rather than overwriting the previous one.
Each version carries its own timestamp, evidence list, and signed receipt, forming an unbroken chain of custody for every claim ever made through the platform.
This design ensures transparency and accountability: if an answer changes, you can always trace exactly why whether it’s new evidence, retraction of an old source, or discovery of a contradiction.

In Practice

Imagine asking, “What are the latest rules on exporting AI models from the UK?”
In April, WebCrux shows one set of sources and quotes.
By November, new legislation has passed. You re-run the same query in Verified mode and instantly see which sources are new, which ones have changed, and which statements are now outdated complete with the audit trail of both answers.

Over time, this forms a personalised timeline of verified knowledge.
You can watch the world’s evidence shift and trust that no matter how facts evolve, your historical record remains intact and provable.

Understanding Bias and the ATAM Framework

Every information system carries the risk of bias sometimes from the data it consumes, sometimes from the algorithms that process it, and often from the people who interpret the results.
The ATAM Framework short for Auth, Trust & Anti-Manipulation is CueCrux’s built-in safeguard against those forces.
It doesn’t hide bias; it detects, labels, and dampens its influence so that users can see why a particular answer leans one way and how much confidence to place in it.

Why Bias Exists

Bias can enter the chain in several ways:

• Source bias: An original article or dataset may reflect a political, commercial, or regional perspective.
• Selection bias: Algorithms can over-favour certain domains or voices, even unintentionally.
• Temporal bias: Older evidence can linger after the world has changed.
• Interaction bias: Feedback loops such as users only trusting what aligns with prior views can distort future recommendations.

ATAM treats these as detectable signals, not invisible flaws.
Instead of pretending neutrality, it quantifies and records them as part of the evidence trail.

How ATAM Works

ATAM operates quietly across every service in the CueCrux platform, analysing both human and machine behaviour through three core layers:

1. Authentication & Identity Integrity

Before trust can exist, identity must be stable.
WebCrux relies on short-lived, cryptographically signed sessions; WatchCrux independently verifies version and origin claims.
The platform detects “sybil” or duplicate accounts, unusual activity spikes, and collusive behaviour across accounts reducing the weight of coordinated or suspicious contributions without blocking legitimate users.

2. Evidence Health & Provenance Checks

ATAM continuously assesses each artefact in the evidence graph:

• Licence & legality: Is the source authorised for use?
• Venue reputation: Is it from a known, reliable publication or dataset?
• Retracted or superseded material: Has this claim been withdrawn or updated elsewhere?
• Independence: Are multiple citations actually quoting the same root source?

Every factor becomes a metric, feeding into WebCrux’s Trust Score for that artefact and its derived answers.
Lower-quality or collusive sources aren’t deleted they’re simply weighted down, and their status is visible to the user.

3. Anti-Manipulation & Transparency Layer

When manipulation is detected for example, a sudden surge of identical citations or a pattern of coordinated linking ATAM logs a trust anomaly.
Affected artefacts remain viewable but are marked with clear reasoning such as “suspected collusion” or “retracted source”.
Users can still inspect and even appeal these findings by supplying counter-evidence, ensuring the system remains open and self-correcting rather than censorious.

A Bias-Aware, Not Bias-Blind Platform

Unlike traditional moderation systems that remove or hide controversial material, ATAM’s goal is traceable fairness.
All trust adjustments are recorded in the provenance ledger with the same rigour as evidence itself.
This means:

• Users can see when an answer has weighted or disputed content.
• Researchers can audit why an artefact was damped or promoted.
• Organisations can demonstrate compliance with neutrality and evidence-handling standards.

How Users Benefit

For end users, ATAM ensures:

• Transparent scoring: Each answer carries a visible trust indicator based on evidence quality, diversity, and contradiction rate.
• No unseen filtering: Contradictory or minority viewpoints are shown with context, not suppressed.
• Adaptive weighting: As new information arrives, trust scores update automatically, reducing the influence of outdated or biased material.

The Philosophy Behind ATAM

ATAM embodies CueCrux’s guiding principle: proof over opinion, context over censorship.
It assumes that disagreement is natural and often informative and that the best remedy for bias isn’t silence, but sunlight.
By encoding fairness directly into its data model, WebCrux can offer answers that remain open, explainable, and resistant to manipulation at scale.

In short: ATAM turns bias from a hidden liability into an observable property of knowledge itself measurable, reviewable, and correctable.

ATAM in Context - Where It Lives in the Stack

ATAM isn’t a separate add-on; it’s embedded into the CueCrux protocol itself.
Every artefact that enters the system whether through Engine retrieval, Factory ingestion, or user upload passes through ATAM’s verification layer before it can influence a result.
From there:

• EngineCrux enforces ATAM’s trust weights during retrieval and ranking.
• WebCrux exposes those weights visually through trust indicators and “Why Trust” panels.
• WatchCrux independently audits anomaly rates and reports them as PASS, WARN, or FAIL in operator findings.
• OpsCrux aggregates these findings across services, surfacing them in the platform’s dashboards.
• InfraCrux preserves the underlying telemetry, ensuring that bias-related metrics remain immutable and reviewable over time.

In the larger architecture, ATAM acts as the connective tissue between evidence and confidence the mechanism that ensures the CROWN trust framework operates on clean, transparent data.
It is what allows the Global Provenance Graph to exist without centralised control: every bias correction, weighting, and retraction is part of the public proof chain, not an invisible algorithmic decision.

Together, ATAM and CROWN form the ethical and technical heart of CueCrux - one safeguarding how truth is measured, the other proving why it can be trusted.

The Bigger Picture: A Provenance-Native Web

Every action in CueCrux contributes to a broader global map of knowledge provenance a Proof Graph that spans regions, industries, and time. Public artefacts strengthen the shared record; private ones remain sealed yet still produce attestable receipts that prove integrity without exposing content.

Together, they form a network of provable knowledge where every answer, dataset, and claim can be placed in context.

The State of the Internet - Signal Lost in the Noise

The modern web is experiencing a crisis of credibility. Information is multiplying faster than it can be verified.

Synthetic articles, AI-generated chatter, and low-effort content farms flood indexes. Search engines now crawl oceans of repetition rather than depth, and even reputable sources struggle to stay visible.

This is the landscape behind what many call the “dead internet” - a web where automated systems mostly talk to other automated systems, recycling the same unverifiable fragments until truth itself blurs.

The CueCrux Response - Verifiable Lineage, Not Volume

CueCrux tackles this problem head-on.

Instead of hoarding data, it curates evidence with lineage. Every artefact must declare its origin, timestamp, and licence before entering the ecosystem.

When an AI model produces a summary, CueCrux forces it to anchor that output to the underlying evidence a process called anchored reasoning. Nothing is ever accepted at face value; everything must have a trace.

This quality-first design turns CueCrux into the web’s immune system: noisy or unreliable sources are automatically weighted down, while strong, verifiable ones reinforce the graph.

Users don’t need to guess which content is genuine they can see the provenance for themselves.

Beyond Caching - Building a Living, Self-Healing Record

Where conventional search engines cache static snapshots of a decaying web, CueCrux builds a living knowledge network.

Each verified artefact strengthens the Proof Graph, while retracted or superseded sources are versioned forward instead of being silently replaced.

The result is a self-healing record one that evolves with the world rather than eroding with it. For AI systems trained on CueCrux data, this means traceability by default.

Every fact, figure, or citation carries a chain of custody, ensuring that generative models remain explainable and ethically grounded rather than opaque and derivative.

A Web That Remembers

This is the essence of the CueCrux vision: a web that doesn’t just serve information, but remembers it immutably, transparently, and verifiably.
A web resilient to manipulation, resistant to decay, and rich enough in provenance to answer the hardest question of the digital age:

How do we know what is real?

CueCrux’s answer is simple, we prove it.

Key Problems WebCrux Solves

Every organisation, researcher, and creator faces the same modern dilemma: information is everywhere, but certainty is scarce.

WebCrux exists to replace assumptions with verifiable truth tackling the most common pain points faced by professionals and everyday users alike.

“I don’t have time to verify this.”

Modern workflows demand speed, but manual verification costs hours.

WebCrux solves this by embedding citations, timestamps, and source previews directly into every answer. Verification is no longer a separate task it’s built in.

Users can hover to inspect the exact line, timestamp, and domain that supports each claim, and one click opens the full artefact for independent confirmation. Whether you’re checking a figure for a report or validating a legal reference, proof is always attached to the claim, not buried behind it.

“I need audit trails for compliance.”

In regulated industries from healthcare and finance to public policy - evidence handling must be traceable and repeatable.

WebCrux introduces Audit Mode, which produces a cryptographically signed receipt for every verified interaction. Each receipt captures the question, sources, model parameters, timestamps, and even counterfactual checks.

These receipts serve as digital audit trails that satisfy internal governance requirements and external regulatory standards, allowing organisations to demonstrate not just what they decided, but how they decided it.

“I can’t risk exposing backend keys.”

Security breaches often occur when client applications expose internal tokens or API credentials.

WebCrux avoids this entirely through a Backend-for-Frontend (BFF) architecture. All sensitive keys, API tokens, and model credentials remain on the server never in the browser or user device.

Requests from the client are proxied securely through the BFF, which injects the necessary credentials only at runtime. This architecture eliminates a major attack surface and keeps data exchanges between WebCrux and the CueCrux Engine confidential, controlled, and auditable.

“Our data is sensitive.”

Many teams handle private datasets, research drafts, or pre-publication content that cannot leave secure environments.

WebCrux uses role-based workspaces, Vault-managed encryption, and Private Stack isolation to keep sensitive data scoped to its rightful owners.

Administrators can control who can view, edit, or re-run specific answers while still benefiting from the shared verification backbone. Even within private deployments, all actions are logged and receipts are generated meaning security and transparency coexist rather than compete.

“We need to trust what the AI tells us.”

As AI-generated content proliferates, so do doubts about its reliability. WebCrux addresses this directly with provenance enforcement and operator oversight.

Every AI-generated answer includes:

• Provenance chains showing which artefacts were used and when.
• Contradiction flags that reveal when credible evidence disagrees.
• Independent audits run by WatchCrux, ensuring that trust metrics, version drift, and bias corrections remain visible to users and administrators alike.

This transforms AI from a black box into a transparent collaborator one whose reasoning can be inspected, challenged, and improved.

With WebCrux, users no longer have to “trust the model”; they can trust the proof.

In Summary

WebCrux redefines what it means to search, verify, and decide in the age of AI. It’s not just a tool for finding answers it’s a framework for proving them.

By blending rigorous provenance with modern design, WebCrux turns uncertainty into a measurable, auditable, and defensible state of knowledge.

Some Example Use Cases

CueCrux exists to restore confidence in information when the cost of being wrong has never been higher.
From researchers to regulators to everyday professionals, people spend more time checking facts than learning from them.
CueCrux changes that not by giving more answers, but by giving answers that prove themselves.

1. Publishing Reliable Research in an Age of Information Saturation

A policy research team is racing to release a paper on AI safety. They have thousands of references, half of which now point to rewritten articles or synthetic “academic” papers generated by bots. The team spends days verifying sources that should have taken minutes.

By running their queries through CueCrux, every cited claim arrives already paired with timestamps, domain provenance, and licence details.
When they export their report, each fact includes a verifiable citation block that readers can inspect directly. CueCrux doesn’t just find references it maintains a permanent record of the evidence used, even if the original page disappears. What once took a week of manual checking becomes a one-click verification step, freeing analysts to think instead of chase links.

2. Making High-Stakes Decisions You Can Defend

An executive team preparing a public statement on new sustainability targets knows that a single inaccuracy could spark regulatory review or shareholder panic. They don’t just need information; they need proof that their figures and forecasts are based on validated data.

CueCrux allows them to run their checks in Audit Mode, generating cryptographically signed receipts for every supporting fact. Each receipt shows where the data came from, when it was last verified, and how it has changed over time.

If a regulator later asks for evidence, the company can replay the receipt and demonstrate the chain of custody for every number quoted. Decisions that once carried legal and reputational risk now come with mathematical accountability confidence that’s defensible under scrutiny.

3. Eliminating Customer Confusion with Proof-Based Answers

A fast-growing SaaS company is overwhelmed by repetitive support tickets. Users copy snippets from outdated forums, quote inaccurate AI answers, and demand fixes for problems that were solved months ago.

Support agents waste hours re-verifying old issues instead of helping new customers.

CueCrux gives the company a single, trust-verified knowledge surface. Every article, update, and how-to is stored as a verifiable artefact with source, version, and timestamp baked in. When customers search for answers, they see results backed by live citations, not guesses.

Agents can reply with confidence, knowing the response is grounded in the most current, proven information.

The result? Faster resolutions, fewer escalations, and a 40-50 % drop in duplicated tickets proof that clarity saves both time and trust.

4. Safe Data Ingestion Without Legal Grey Areas

A marketing analytics firm routinely collects public datasets for trend reports.
After a costly legal scare unknowingly reusing a restricted dataset - they need to ensure every file they analyse is legitimately sourced and properly licensed.

With CueCrux, data collection runs through a policy-aware verification layer. Each dataset is automatically checked for licence type, jurisdiction, and provenance before it’s added to the firm’s reports. If a source later changes its terms or is retracted, the team receives an alert along with an updated provenance record.

No lawyers, no manual audit spreadsheets just datasets that come with their own proof of permission. What used to be a compliance bottleneck becomes a seamless part of everyday workflow.

5. Restoring Confidence in AI-Generated Content

A content agency experimenting with AI assistants faces a growing problem: their writers are faster, but they can’t tell which parts of the text are genuinely sourced and which are model hallucinations. Editors spend more time fact-checking than writing.

CueCrux solves this by requiring every AI-generated statement to carry source evidence and contradiction checks before it’s published.

Writers can hover over each claim to view its supporting citations, and editors can export a proof report showing every artefact used in generation. Instead of post-hoc verification, proof becomes part of the creative process ensuring that speed never sacrifices credibility. The agency now delivers content twice as fast with measurable accuracy, turning AI from a liability into a trusted collaborator.

In Summary

Across industries, the pattern is the same: people no longer struggle to find information they struggle to trust it. CueCrux gives back the hours lost to re-checking, the confidence lost to uncertainty, and the credibility lost to misinformation.

It turns data chaos into a network of verifiable truth, where every answer saves time, avoids risk, and earns belief.

The Crux Economy - Proof, Not Points

Every action inside CueCrux creates or consumes value.

To measure that value fairly, we built Crux (CRUX = Credit Resource Unit eXchange): internal participation credits that power the entire ecosystem.

Crux is not money, a token, or a cryptocurrency. It is a trust-metering system that rewards the creation of verifiable knowledge and pays for the computational cost of maintaining it.
CRUX is pegged to real operating costs via a Proof Cost Index and reviewed quarterly.

What Crux Represents

Crux sits at the intersection of trust and compute.

When you contribute something provably useful a verified dataset, a well-cited artefact, or a reproducible audit you earn Crux. When you use the network’s compute power or storage to process, verify, or archive new data, you spend it.

This balance keeps CueCrux self-regulating: the more trustworthy the content in the system, the richer the community becomes in Crux; the more users consume verification or storage, the more they must replenish the network through contribution.

Why a Proof Credit, Not a Token

We designed Crux to encourage collaboration, not speculation.

It can’t be traded or cashed out; it exists purely to balance the cost of truth.

Crux keeps the ecosystem healthy by ensuring that what’s valuable isn’t volume, but verifiability.

Earning Crux - Quality Over Quantity

You earn Crux by making the network smarter, cleaner, and more reliable.
The system measures not just that you contributed, but how much trust your contribution added.

Each reward is weighted by Trust Score, a composite metric derived from:

• Evidence quality – freshness, citation independence, licence health.
• Reputation of sources – historically accurate domains earn higher weight.
• Reuse impact – how often your artefacts appear in verified answers.
• Contradiction rate – lower contradiction means higher reliability.

In other words, 10 KB of verified truth is worth more than 10 MB of noise.

Spending Crux - Paying the Cost of Proof

Every proof has a cost: computation, storage, and bandwidth.
Crux covers those costs transparently so users can see exactly what their certainty consumes.

If you run out of Crux, you can top up through contribution (adding verified artefacts) or by purchasing compute time from CueCrux.

However, purchase only covers the energy cost reputation and trust still have to be earned.

Compute as Contribution

One of the most powerful aspects of the Crux economy is that computation itself can earn Crux. Users or organisations can lend unused compute power to CueCrux’s verification pool. When their machines help validate artefacts or replay receipts, they receive Crux proportional to the quality and trust impact of their processing.

For example:

• A local node re-verifies 1 GB of government documents, confirming all licences and timestamps → earns ≈ 20 Crux.
• A GPU worker performs counterfactual analysis on AI-generated claims → earns ≈ 15 Crux for improving contradiction coverage.

This makes CueCrux a participatory compute network, where verifying truth fuels the system more efficiently than simply storing it.

The Role of Trust Scores

Every participant human or machine has a dynamic Trust Score. It rises with consistent, verifiable contributions and falls with failed audits, contradictions, or licence violations.

Higher Trust Scores amplify Crux rewards and unlock privileges such as:

• Faster verification queues.
• Larger free storage tiers.
• Early access to experimental tools.

This encourages sustainable, evidence-driven behaviour across the network.

Transparency and Fairness

All Crux transactions earn, spend, decay, and reuse are logged in the Provenance Ledger.

No one can alter their balance without leaving a trace.

The system publishes anonymised metrics such as total Crux minted, average trust scores, and reuse ratios, ensuring the economy stays transparent and inflation-controlled.

Why It Matters

The internet currently rewards attention, not accuracy.

CueCrux reverses that logic. By tying value to verifiable contribution, Crux transforms accuracy into an incentive. The more you help the world know with certainty, the more capability you earn to keep verifying it.

It’s an economy where truth generates its own currency of trust a sustainable cycle of proof that keeps knowledge alive, accurate, and fair.

Reliability & Observability

Trust doesn’t stop at data it extends to the platform itself.
CueCrux is engineered to be as reliable as the answers it delivers.
Every service, from search to provenance to storage, is continuously measured, audited, and self-correcting.

Consistent Performance You Can Feel

Speed and reliability aren’t marketing claims they’re measured targets.
CueCrux tracks how quickly the system responds to user requests at scale, using what engineers call the p95 benchmark meaning 95% of all requests complete within this time or faster.

Our current targets are:

• Verified mode: Most responses arrive in under 160 milliseconds, roughly the blink of an eye.
• Audit mode: Even with full provenance checks and counter-evidence scanning, responses typically return within 400 milliseconds.
• End-to-end answers: From question to fully cited result, users receive their answer in around 2 seconds, even under heavy load.

For the average user, this means CueCrux feels instant and consistent whether it’s one query or ten thousand.

Operational Transparency and Monitoring

All services within CueCrux are continuously monitored and observable through an integrated metrics pipeline.

• InfraCrux collects performance data from every subsystem network latency, error rates, storage health, and cost.
• Prometheus metrics feed directly into Grafana dashboards, giving engineers live insight into the system’s heartbeat.
• WatchCrux, the independent operator service, runs automated audits and publishes its findings directly into OpsCrux, where anomalies are flagged within minutes.

Every component is visible, every performance trend is measurable, and every change leaves an audit trail.

Proven Resilience and Disaster Recovery

CueCrux assumes that things can go wrong and designs for it.
The platform follows a self-healing resilience model, meaning that if one component slows or fails, the rest of the system isolates and recovers automatically.

Key practices include:

• Circuit-breaker logic If a service becomes unresponsive, CueCrux stops sending requests to it temporarily, preventing a ripple effect of failures.
• Automatic retries with backoff The system reattempts requests intelligently, avoiding overload.
• Version-drift detection If any service runs an outdated or mismatched version, OpsCrux raises an alert within 60 seconds.
• Canary releases and safe rollbacks New updates are released gradually to a small percentage of users, monitored for stability, and automatically rolled back if issues arise.

Together, these ensure that CueCrux can update itself, defend itself, and recover gracefully with no data loss and minimal downtime.

Verified Backups and Recovery Assurance

Every day, the system quietly proves that your data is safe.
All key databases including the Provenance Ledger and user artefacts are backed up continuously with Point-in-Time Recovery (PITR) tested to within 30 minutes.
Monthly restore drills confirm that these backups are valid and can be restored in real-world scenarios.
If a server fails or data corruption is detected, the system can roll back precisely to the last verified state.

This approach transforms backups from a checkbox into a living safety net one that’s proven, not promised.

Observability in Plain Sight

Users don’t see dashboards but they benefit from them.
OpsCrux provides real-time operational intelligence to the CueCrux team:

• Tracking system health, uptime, and latency across all regions.
• Monitoring cost and compute efficiency so Crux credit pricing stays fair.
• Watching for anomalies like sudden contradiction spikes or provenance errors, and triggering automated investigations via WatchCrux.

Every operational event from a backup restore to a software release - is logged, time-stamped, and independently verifiable.
This makes the platform as auditable as the knowledge it serves.

Why Reliability Matters

CueCrux’s mission is to make truth measurable but that means our infrastructure has to be measurable too.
A proof-based ecosystem cannot rely on trust alone; it must continuously demonstrate its own integrity.
By combining live observability, proactive auditing, and verifiable recovery, CueCrux ensures that users never have to wonder if the system is accurate, secure, or available.
It simply is and we can prove it.

GDPR and Global Data Protection Alignment

CueCrux was designed from day one to comply with the world’s strongest privacy and data-protection standards not as an afterthought, but as a core part of the architecture.
The platform aligns with EU GDPR, UK Data Protection Act 2018, and equivalent US state privacy frameworks (including CCPA/CPRA).
Every region, tenant, and feature is built on the same privacy-by-design foundation.

Privacy by Design, Not Retrofit

Rather than bolt compliance on after launch, CueCrux embeds it directly into the data model and workflow logic:

• Data minimisation: Only the minimum personal data required for authentication and audit logging is collected.
• Purpose limitation: Each artefact and event has an explicit purpose field recorded in the Provenance Ledger; data cannot be repurposed without creating a new, signed entry.
• Storage limitation: Artefacts, user data, and logs follow defined retention windows deletions cascade automatically once obligations expire.
• Access limitation: Role-based controls and Vault-managed encryption keep sensitive content confined to authorised users and tenants.

These principles mirror Articles 5 and 25 of GDPR and equivalent UK and US privacy clauses.

DSAR (Data-Subject Access Request) Support

Any verified user can request a copy of their stored personal data at any time.
CueCrux automates the Data-Subject Access Request (DSAR) process through its privacy portal:

1. The user submits a request through the WebCrux interface or API.
2. The platform gathers all relevant data accounts, receipts, logs - from the Provenance Ledger and operational databases.
3. A structured export (JSON or PDF) is generated within the legal timeframe (≤ 30 days EU/UK; ≤ 45 days US).
4. All exports are cryptographically signed to confirm integrity and timestamped for audit purposes.

This ensures transparency and accountability without manual intervention.

Erasure & Right-to-Be-Forgotten Workflows

When a user or organisation requests deletion, CueCrux performs verified erasure across all systems:

• Personal identifiers are removed or anonymised.
• Associated artefacts are re-linked to pseudonymised placeholders to maintain the structural integrity of the ledger.
• Backups containing deleted data are flagged for secure destruction within the next rotation cycle.
• Proof-of-erasure receipts are generated and stored for audit.

This process satisfies GDPR Articles 16–19, UK DPA 2018 Schedule 2, and relevant US state equivalents, ensuring compliance while preserving the integrity of shared evidence graphs.

Regional Data Residency and Isolation

CueCrux operates with region-scoped data planes:

• EU and UK data are processed and stored exclusively within compliant regional data centres.
• Cross-border transfers require explicit consent and standard contractual clauses (SCCs).
• Enterprise tenants using Private Stack instances can enforce their own jurisdictional boundaries and retention policies.

This ensures lawful processing under EU/UK transfer rules and simplifies compliance for multinational organisations.

Independent Oversight and Auditing

• WatchCrux performs independent audits of privacy metrics, retention compliance, and key-rotation schedules.
• OpsCrux exposes compliance dashboards to administrators, showing DSAR volumes, erasure requests, and audit outcomes.
• InfraCrux maintains immutable logs for all privacy events, secured under Vault encryption.

External audits can verify that privacy controls are functioning as designed, giving regulators and partners tangible proof of compliance.

Alignment With Emerging US and Global Standards

CueCrux’s privacy framework already aligns with current and forthcoming US data-protection laws, including CCPA/CPRA (California), CPA (Colorado), and VCDPA (Virginia), as well as NIST Privacy Framework v1.0 guidelines.
This means global users enjoy the same privacy guarantees as those within the EU and UK, without waiting for jurisdiction-specific retrofits.

Why It Matters

In a world where AI and data services increasingly blur the line between private and public information, CueCrux treats privacy as a verifiable contract, not a policy page.
Every consent, access, and deletion event is recorded in the same immutable provenance chain that powers the rest of the platform making privacy itself provable.

In short: CueCrux doesn’t just comply with data-protection law - it demonstrates compliance automatically, with receipts you can trust.

Compliance Map (EU | UK | US | Global)

SDKCrux - Typed Clients for Proof, Not Just Payloads

What SDKCrux Is

SDKCrux is the official TypeScript SDK for developers building on the CueCrux protocol.
It provides strongly-typed access to answers, provenance records, receipts, and trust metrics letting developers display and verify CueCrux data safely within their own applications.

Where a normal SDK simply fetches results, SDKCrux ensures every response can be verified, traced, and reproduced.

What’s Included

• Typed data models (DTOs) for:
  • Answers and Search responses
  • Artefacts and MiSES (QUORUM) groups
  • CROWN snapshots and Receipts
  • Trust metrics (domain diversity, freshness, contradiction rate)
• Provenance helpers
  • verifyReceipt() – validates ed25519 signatures and BLAKE3 hashes
  • quoteSpanCheck() – ensures quoted text matches the captured artefact
  • licenceGuard() – identifies public vs restricted data for compliance
• Budget & retry utilities
  • withBudget({ ms, crux }) – respects C³ cost envelopes and latency targets
  • Automatic exponential back-off and jittered retries
• Development support
  • Mock fixtures for offline testing (via MSW)
  • “Why Trust” UI components (headless) for React and Vue
  • Full OpenAPI client generation for public endpoints

  Example - Quick Verification Flow

import { Answers, Receipts, trustBadge } from '@cuecrux/sdk';

const res = await Answers.ask({
  q: 'Latest EU AI Act amendments 2025',
  mode: 'verified',
  k: 5
});

// Verify the cryptographic receipt
const valid = await Receipts.verify(res.crown.receiptId);

// Display a simple trust badge in UI
const badge = trustBadge(res.trust);
console.log(badge.label); // “High Trust - 2 independent sources”

This minimal snippet shows the SDK handling everything automatically: query routing, provenance validation, and trust scoring all within a few lines.

Where SDKCrux Is Used

Product UIs: Embed live provenance directly into dashboards, reports, or CMS pages.

Compliance Workflows: Validate stored receipts automatically and flag any expired or revoked artefacts.

Editorial & Publishing Tools: Check outbound content for missing citations, and attach cueproof receipts for readers to inspect.

Enterprise Intranets & Wikis: Display up-to-date trust scores beside internal documentation, helping teams separate legacy facts from verified ones.

Why SDKCrux Matters

SDKCrux moves proof from the backend into the hands of the user. It allows any developer from SaaS engineer to academic publisher - to give their users visible, reproducible confidence in the information they consume. In a web where AI outputs are easy to fabricate, SDKCrux makes every trusted answer provable at source.

Agent Access & API Integration

As the web evolves, machines will consume and verify information as often as humans do.
CueCrux treats agents autonomous AI models, orchestration layers, and external applications as first-class citizens of the proof network.
Every agent call obeys the same laws of evidence, cost, and provenance as human queries, ensuring that automation doesn’t dilute trust but amplifies it.

Unified API for Humans and Machines

CueCrux exposes a single proof-aware API built for both interactive use and machine orchestration.
Agents connect using lightweight service credentials and can operate fully programmatically:

• POST /v1/answers: ask questions with optional mode (light, verified, audit)
• GET /v1/receipts/:id: retrieve and replay proof records
• POST /v1/search: perform structured provenance searches
• GET /v1/artefacts/:hash: resolve any artefact from the ledger
• POST /v1/agent/usage: record computational or verification work (for Crux rewards)

All endpoints accept either CBOR (Concise Binary Object Representation) or JSON, but agents are encouraged to use CBOR for compactness and performance.
CBOR payloads are smaller, faster to parse, and deterministic ideal for low-latency machine-to-machine communication.

{
  "q": "Latest AI Act enforcement guidelines",
  "mode": "verified",
  "budget": { "crux": 2 },
  "agent_id": "partner.openai.gpt5",
  "response": {
    "summary": "...",
    "mises": [ ... ],
    "crown_receipt": "ed25519:..."
    }
}

Agents can cache receipts locally and later re-verify them offline against the CueCrux ledger.

Paying for Search & Verification with Crux

Every agent interaction consumes or earns Crux, mirroring human use.
This turns proof into a self-funding ecosystem where machines that verify knowledge help sustain the network.

Agents maintain their own Crux wallets and can top-up or earn credits based on the trust impact of their work.
When multiple agents collaborate - for instance, a planner agent asking and a verifier agent confirming - Crux credits flow between them automatically under C³ budgets.

Example - AI Assistant Using CueCrux for Verified Answers

A conversational model wants to ground its responses in verifiable data instead of scraped text.
It calls the CueCrux API using CBOR payloads:

1. Sends a POST /v1/answers query in verified mode (cost: 1 Crux).
2. Receives a compact CBOR response containing citations and a CROWN receipt.
3. Extracts the artefact hashes, validates them locally, and embeds the citation list in its own reply to the user.
4. Optionally submits a POST /v1/agent/usage report for audit rewards (earn +2 Crux).

The agent never touches raw credentials or internal keys all authentication occurs through scoped service JWTs issued via Vault Transit.
Every exchange produces verifiable evidence that can later be inspected by humans or other systems.

Reciprocal Trust - Agents as Verifiers

In the long run, agents won’t just consume knowledge; they’ll maintain it.
By replaying receipts, cross-checking artefacts, or contributing contradiction scans, autonomous verifiers help keep the Proof Graph healthy.
Each machine that validates an artefact earns Crux proportional to its contribution to global trust.

Imagine:

• A legal-analysis bot verifying the freshness and licence status of case-law artefacts.
• A science-domain model re-checking citations across open-access journals nightly.
• An enterprise assistant running scheduled audits on corporate datasets and earning Crux for maintaining compliance.

CueCrux becomes a trust fabric equally usable by people and machines a neutral protocol where both speak the same proof language.

Why This Matters

The future of the internet isn’t human-only or AI-only it’s co-verified.
CueCrux ensures that when autonomous systems ask, store, or share knowledge, they do so with the same rigour as their human creators.
Every packet carries provenance; every model has accountability; every fact leaves a receipt.
That’s how we keep automation honest and how proof stays universal.

CBOR by Default - Efficiency Meets Integrity

CBOR replaces JSON for agent traffic because:

• Smaller footprint: typical 40–60% reduction in payload size.
• Faster serialisation - consistent binary encoding for real-time use.
• Deterministic hashes: artefact digests computed directly on the binary form, preventing whitespace or encoding drift.
• Integrity-first design: each CBOR packet includes an embedded hash map of artefact references, making tampering immediately detectable.

Example (CBOR structure overview):

The payload schema is the same as JSON; CBOR is simply the deterministic binary encoding of that structure (see the example payload above).

Agents can cache receipts locally and later re-verify them offline against the CueCrux ledger.

Getting Started

1. Sign in to your WebCrux account.
2. Ask a question start with Verified mode for best balance of speed and assurance.
3. Review the answer and open the Why Trust panel to inspect citations.
4. Save or share answers to a workspace.
5. Re-run queries later for updated citations or Audit-mode verification.

Glossary

How to Read a CueCrux Receipt

Every verified or audited answer in CueCrux produces a Receipt a compact, cryptographically signed document that proves how that answer was built.
Think of it as the chain of custody for truth: a digital fingerprint showing where evidence came from, how it was processed, and what the outcome means.

A CueCrux Receipt can be opened by any user or agent. You don’t need specialist tools just a standard verification library or the built-in Why Trust viewer.
The structure is consistent across all modes (Verified or Audit) so that humans and machines can both read it easily.

Example Receipt (simplified)

{
  "receipt_id": "recp_2f5a9f3e7b",
  "query": "What are the latest EU AI Act amendments?",
  "mode": "verified",
  "issued_at": "2025-11-17T09:42:33Z",
  "engine_version": "1.1.3",
  "model": "gpt-5",
  "crown": {
    "counterfactual_checked": true,
    "reputation_weighting": "domain+licence",
    "origin_signatures": ["ed25519:…"],
    "workflow_id": "crown_run_895d"
  },
  "mises": [
    {
      "claim": "The AI Act introduces transparency obligations for general-purpose models.",
      "artefacts": [
        {
          "hash": "blake3:53a4c…",
          "domain": "eur-lex.europa.eu",
          "timestamp": "2025-10-20T12:01:45Z",
          "licence": "OGL-3.0",
          "reputation": 0.98
        },
        {
          "hash": "blake3:8aa7b…",
          "domain": "consilium.europa.eu",
          "timestamp": "2025-10-21T09:15:02Z",
          "licence": "CC-BY-4.0",
          "reputation": 0.94
        }
      ]
    }
  ],
  "trust": {
    "score": 0.96,
    "domains": 2,
    "contradictions": 0,
    "freshness_days": 28
  },
  "signature": "ed25519:DEADBEEF…",
  "crux_cost": 1
}

Related Pages

• Technical Overview
• Trust Model
• Forward Vision
• API Reference

Last updated: 12 January 2026