AML Runbook
Guide
Operational runbook for AML/KYC, payout reviews, and WatchCrux alerting.
Read the full card →The docs
The deep material: architecture, receipt anatomy, trust scoring, quorum, compliance, and the defensibility & legal FAQ. Product how-to and marketing FAQ live on cuecrux.com.
Looking for quickstarts, the MCP and auth contracts, or the API reference? Those are the product documentation and live at /docs.
Don't trust the docs
The fastest way to calibrate how much of this documentation to believe is to break the integrity claim yourself, offline, in about a minute. Export a store, verify the bundle with no network, flip one byte, and watch the verifier refuse it:
corecruxctl context export --data-dir <dir> --out ./bundle corecruxctl context verify ./bundle --json # ok=true, offline bash scripts/demo-receipt-tamper.sh # flips a byte, asserts the catch
Every guide below assumes that footing: claims link to source, commands are copy-paste runnable, and anything gated by tier says so explicitly.
Build on the daemon
Thirteen chapters on extending corecruxd: packs, tools, sandboxed modules, connectors, and the signed rail you ship them on. Every claim carries a file:line reference into the Crux repository, and where a surface is declared but has no runtime, it says so.
Guide
Operational runbook for AML/KYC, payout reviews, and WatchCrux alerting.
Read the full card →Guide
External developer guide for calling WebCrux APIs without access to internal networks.
Read the full card →Guide
The building blocks of proof what an artefact is, how it’s created, what it contains, and how it’s scored (plain‑English).
Read the full card →Guide
What a receipt proves, how to verify without code, export trust reports, and keep an audit trail.
Read the full card →Guide
Why disagreement is shown, how it affects trust, and how to proceed.
Read the full card →Guide
Build against the CueCrux platform from a developer’s perspective. Learn when to call the WebCrux API versus the Engine SDK, how to bootstrap your client, and how to keep secrets safe.
Read the full card →Guide
CueCrux’s evidence-driven upgrade system for autonomous agents.
Read the full card →Guide
What common licences mean, how CueCrux labels them, and what you can do with cited evidence.
Read the full card →Guide
Visual primer on the CueCrux planes, their data structures, and how a query flows through them.
Read the full card →Guide
Story-driven explainer of how artefacts are linked, stored, and audited for a single query.
Read the full card →Guide
What a CROWN receipt contains and how to verify it.
Read the full card →Guide
Structural overview, protocol flow, and layered system design for WebCrux.
Read the full card →Guide
How WatchCrux PASS/WARN/FAIL audits flow into WebCrux once the Findings surface is enabled.
Read the full card →Guide
Explainable signals used to assess domains, artefacts, and users without revealing tunable weights.
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
AI systems that cannot prove what they do will face: - regulatory intervention, - enterprise procurement barriers, - escalating liability, - and market…
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
CueCrux answers are not ephemeral text outputs they are *verifiable artefacts*.
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
CueCrux builds answers on QUORUM (MiSES) (Quorum of Unified Observations and Referenced Underlying Material), selecting MiSES (Minimal Evidence Sets) per claim.
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
Every artefact in CueCrux is: - hash-fingerprinted (BLAKE3), - signature-backed, - licence-annotated, - jurisdiction-tagged, - traced through an app…
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
CueCrux is not a crawler.
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
CRUX credits anchor platform economics to *actual compute cost*: - Verified query = 1 CRUX - Audit query = 5 CRUX - Hot storage = 1 CRUX / 10MB / mo -…
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
CueCrux already implements: - provenance transparency, - evidence traceability, - risk classification, - contradiction detection, - audit logs, - D…
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
CueCrux is a federation of interoperable planes: - EngineCrux: Proof & Retrieval - WebCrux: Identity & BFF isolation - FactoryCrux: Ingestion & Licensin…
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
WatchCrux provides: - continuous health polling, - version drift detection, - receipt validation, - SLO monitoring, - metrics ingestion, - audit ar…
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
SDKCrux enforces: - strict DTO schemas, - semantic versioning, - compatibility requirements, - signature verification, - receipt replay tooling, - …
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
Private Stack gives enterprises: - tenant-isolated queues, storage, databases - tenant-scoped signing keys - private ingestion and agents - custom rank…
Read the full card →Defensibility FAQ · Why This Platform Wins Long-Term
CRUX rewards: - link submission - first-capture evidence - dedupe and curation - reuse in QUORUM (MiSES) - audit verifications ##### Why this is a…
Read the full card →Legal & Compliance FAQ
CueCrux builds every answer on evidence with provenance, timestamps, and content hashes.
Read the full card →Legal & Compliance FAQ
CueCrux treats personal data with strict separation rules.
Read the full card →Legal & Compliance FAQ
FactoryCrux enforces robots.txt, X-Robots-Tag, and licence metadata before ingestion.
Read the full card →Legal & Compliance FAQ
CueCrux performs nightly retraction syncs (e.g., Crossref, Retraction Watch).
Read the full card →Legal & Compliance FAQ
CueCrux is designed for provable compliance.
Read the full card →Legal & Compliance FAQ
The ATAM (Auth, Trust & Anti-Manipulation) system detects: - collusive citation loops - coordinated manipulation attempts - prompt-injection threats - …
Read the full card →Legal & Compliance FAQ
CueCrux does not assume that uploaded documents are correct; it treats all submissions as claims that must be verified through retrieval and evidence checks.
Read the full card →Legal & Compliance FAQ
CueCrux is designed to avoid reproducing unverified or defamatory statements.
Read the full card →Legal & Compliance FAQ
Many users upload incorrect or outdated information unintentionally.
Read the full card →Legal & Compliance FAQ
CueCrux integrates multiple defences from the ATAM (Auth, Trust & Anti-Manipulation) framework: - Similarity detection prevents mass-uploading of near-duplic…
Read the full card →Legal & Compliance FAQ
CueCrux benefits from strong architectural and procedural defences: - It does not publish uploads directly; it processes them internally.
Read the full card →Legal & Compliance FAQ
Bias concerns typically arise when AI models produce ungrounded or one-sided content.
Read the full card →Legal & Compliance FAQ
CueCrux’s trust architecture is designed to satisfy emerging AI regulatory frameworks (AI Act, DSA, FTC guidance).
Read the full card →Legal & Compliance FAQ
No system is perfect, but CueCrux’s layered safety stack makes harmful outputs extremely unlikely.
Read the full card →Legal & Compliance FAQ
CueCrux is architecturally aligned with legal “safe harbour” principles because: - It does not publish user uploads; it processes them internally.
Read the full card →Legal & Compliance FAQ
CueCrux maintains: - append-only provenance ledgers, - CROWN receipts, - WatchCrux audit reports, - OpsCrux system events, - task and ingestion logs,…
Read the full card →Legal & Compliance FAQ
Because CueCrux does not *claim correctness*; it claims provenance.
Read the full card →