Skip to content
Black Atlas — Powered by KRYOS V6

Reference

The Forty-Slot Engine Registry

Each analysis engine is listed with the status recorded in the source corpus and the guard that applies to it. Status is reproduced as written, including where it is unfavourable.

  1. 01

    Direct Prompt Injection Simulation

    Restricted to isolated synthetic validation; qualified human review mandatory.

    STATUS NOT EXPOSED IN ACCESSIBLE CORPUS
  2. 02

    Indirect Prompt Injection via Context Manipulation

    Restricted to isolated synthetic validation; qualified human review mandatory.

    STATUS NOT EXPOSED IN ACCESSIBLE CORPUS
  3. 03

    Memory Poisoning Simulation

    Restricted to isolated synthetic validation; qualified human review mandatory.

    STATUS NOT EXPOSED IN ACCESSIBLE CORPUS
  4. 04

    Authoritative engine name unavailable

    Do not infer or invent. Retrieve the current authoritative KRYOS V6 source before activation.

    BLOCKED - SOURCE GAP
  5. 05

    Authoritative engine name unavailable

    Do not infer or invent. Retrieve the current authoritative KRYOS V6 source before activation.

    BLOCKED - SOURCE GAP
  6. 06

    Authoritative engine name unavailable

    Do not infer or invent. Retrieve the current authoritative KRYOS V6 source before activation.

    BLOCKED - SOURCE GAP
  7. 07

    Authoritative engine name unavailable

    Do not infer or invent. Retrieve the current authoritative KRYOS V6 source before activation.

    BLOCKED - SOURCE GAP
  8. 08

    Authoritative engine name unavailable

    Do not infer or invent. Retrieve the current authoritative KRYOS V6 source before activation.

    BLOCKED - SOURCE GAP
  9. 09

    Authoritative engine name unavailable

    Do not infer or invent. Retrieve the current authoritative KRYOS V6 source before activation.

    BLOCKED - SOURCE GAP
  10. 10

    Authoritative engine name unavailable

    Do not infer or invent. Retrieve the current authoritative KRYOS V6 source before activation.

    BLOCKED - SOURCE GAP
  11. 11

    Retrieval Manipulation Detection and Analysis

    Use reproducible query, corpus, and retrieval configuration evidence.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  12. 12

    Corpus Poisoning and Data Integrity Analysis

    Use synthetic artifacts and isolate the production corpus.

    OPERATIONAL, REQUIRES SPECIALIZED LABORATORY
  13. 13

    Tool Misuse and Abuse Simulation

    Use stubs, dry runs, synthetic accounts, and approved sandboxes.

    CONFIGURED BUT UNVALIDATED, AUTHORIZED TEST ENVIRONMENT ONLY
  14. 14

    Excessive Agency and Unbounded Action Analysis

    No autonomous production changes based solely on the analysis.

    RESEARCH ONLY, REQUIRES QUALIFIED HUMAN REVIEW
  15. 15

    Privilege Escalation Path Discovery

    No credential theft, persistence, or unapproved privilege change.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  16. 16

    Cross-Tenant Leakage and Isolation Failure Analysis

    Use synthetic tenants or explicitly approved data and accounts.

    OPERATIONAL, REQUIRES INTEGRATION
  17. 17

    Unauthorized Federation and Trust Boundary Testing

    Do not treat as operational until specified, implemented, and validated.

    PROPOSED, REQUIRES SPECIALIZED LABORATORY
  18. 18

    Agent Identity Confusion and Impersonation Analysis

    Use synthetic identities and bounded identity-binding tests.

    REQUIRES QUALIFIED HUMAN REVIEW, AUTHORIZED TEST ENVIRONMENT ONLY
  19. 19

    Correlated Reasoning Failure Detection

    Treat model disagreement and correlated failure as uncertainty signals.

    RESEARCH ONLY, REQUIRES QUALIFIED HUMAN REVIEW
  20. 20

    Consensus Gaming and Manipulation Analysis

    Preserve dissent; do not manufacture consensus.

    REQUIRES INTEGRATION, REQUIRES QUALIFIED HUMAN REVIEW
  21. 21

    Arbiter Manipulation Detection and Analysis

    Use independent adjudication evidence and role separation.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  22. 22

    Evaluator Manipulation and Output Bias Analysis

    Document evaluator criteria, conflicts, and performance limits.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  23. 23

    Source-Authority Substitution Testing

    Use authority hierarchy and currentness checks.

    RESEARCH ONLY, REQUIRES QUALIFIED HUMAN REVIEW
  24. 24

    Citation Fabrication and Hallucination Analysis

    Trace every material claim to a retrievable source segment.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  25. 25

    Provenance Tampering and Traceability Testing

    Validate lineage across source, transformation, retrieval, and output.

    REQUIRES INTEGRATION, REQUIRES QUALIFIED HUMAN REVIEW
  26. 26

    Stale-Source Retrieval and Data Freshness Analysis

    Define effective dates, supersession, and freshness thresholds.

    OPERATIONAL, REQUIRES SPECIALIZED LABORATORY
  27. 27

    Model Substitution and Unauthorized Model Swap Testing

    Use approved registries and synthetic deployment checks.

    CONFIGURED BUT UNVALIDATED, AUTHORIZED TEST ENVIRONMENT ONLY
  28. 28

    Configuration Drift and Unauthorized Change Detection

    Reconcile changes against approved baselines and change records.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  29. 29

    Race Condition and Timing Failure Simulation

    No disruptive concurrency testing in production.

    REQUIRES SPECIALIZED LABORATORY, REQUIRES QUALIFIED HUMAN REVIEW
  30. 30

    State Corruption and Integrity Failure Analysis

    Protect production state; use snapshots, fixtures, or isolated environments.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  31. 31

    Retry/Duplicate-Action Failure Analysis

    Validate idempotency and prevent real duplicate external actions.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  32. 32

    Connector Compromise and Supply Chain Injection Testing

    Use owned, isolated connector replicas and synthetic artifacts.

    REQUIRES INTEGRATION, REQUIRES SPECIALIZED LABORATORY
  33. 33

    Secret Exposure and Credential Leakage Analysis

    Redact full secrets and coordinate authorized containment.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  34. 34

    Unsafe Fallback and Default-Path Failure Analysis

    Do not trigger unsafe production fallback behavior.

    RESEARCH ONLY, REQUIRES QUALIFIED HUMAN REVIEW
  35. 35

    Monitoring Blindness and Observability Failure Detection

    Validate path and evidence coverage, not logging volume alone.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  36. 36

    Rollback Failure and Recovery Path Analysis

    Require backups, stop conditions, and safe recovery environments.

    OPERATIONAL, REQUIRES QUALIFIED HUMAN REVIEW
  37. 37

    Advanced Cross-Domain Data Flow and Policy Violation Detection

    Normalize domains without erasing policy differences.

    REQUIRES INTEGRATION, REQUIRES QUALIFIED HUMAN REVIEW
  38. 38

    Advanced Cross-Domain Privilege and Trust Escalation Testing

    Use synthetic cross-domain identities and bounded trust tests.

    REQUIRES SPECIALIZED LABORATORY, REQUIRES QUALIFIED HUMAN REVIEW
  39. 39

    Advanced Cross-Domain Configuration and Policy Drift Detection

    Do not elevate confidence without independent evidence.

    CONFIGURED BUT UNVALIDATED, AUTHORIZED TEST ENVIRONMENT ONLY
  40. 40

    Advanced Cross-Domain Resilience and Systemic Weakness Simulation

    Use for hypothesis and stress testing, not automated production decisions.

    RESEARCH ONLY, REQUIRES SPECIALIZED LABORATORY