TFTHREATFADE
ProductDetectionHow it worksIntegrationsResearchValidationSecurity & TrustDocsPlaygroundPricingEnterprise
GitHub
ProductDetectionHow it worksIntegrationsResearchValidationSecurity & TrustDocsPlaygroundPricingEnterprise
HomeValidation / evidence

Evidence with a boundary, not a headline.

ThreatFade separates deterministic repository validation from external validation. Every published result identifies its method, scope, limitations and reproducibility path.

v0.9.0-devDevelopment line

The authoritative evidence boundary is maintained in the engine repository. This page is a synchronized presentation surface; it does not create new validation evidence.

Read the validation boundary →
Evidence records

What has actually been evaluated?

TF-SYNTH-V2

Synthetic detection-quality regression

VALIDATED
Category
Deterministic Benchmark
Methodology
benchmarks/benchmark.py runs 100 deterministic seeds for each of five declared scenarios.
Dataset / sample
synthetic-scenario-v2 · 500 cases total
Result boundary
Regression gate requires all seeded malicious scenarios to remain detectable and the declared benign scenario to remain free of false positives.
Limitations
Synthetic regression is not real-world, customer-scale, independent or production validation.
Reproducibility
Run python benchmarks/benchmark.py from the repository root with the documented dependencies.

TF-EXT-V0

External labeled-traffic validation

PLANNED
Category
Independent Validation
Methodology
Frozen protocol requiring provenance-controlled labeled external traffic, leakage controls and independent evaluation.
Dataset / sample
Not yet attached
Result boundary
No external result is claimed.
Limitations
Real-world labeled-corpus and independent evaluation remain outstanding.
Reproducibility
Protocol and data contract are documented; execution requires the external corpus/evaluator.
Evidence taxonomy

Implementation is not the same as validation.

IMPLEMENTED

Capability exists in the current codebase.

TESTED

Automated or reproducible tests exercise the capability.

VALIDATED

Documented evaluation evidence exists for a defined scope.

EXPERIMENTAL

Research-stage capability; not authoritative for production decisions.

PLANNED

Not currently implemented.

ADAPTER

Interoperability is provided through an adapter rather than a native vendor integration.

EXTERNAL-DEPENDENCY

Operation depends on an external provider or deployment boundary.

What this page does not claim

  • • No independent detection validation.
  • • No customer validation claim.
  • • No universal false-positive guarantee.
  • • No independently reproduced throughput claim.
  • • No SOC 2 / ISO certification claim.
  • • No third-party penetration-test claim.
THREATFADE / TINLANCE LIMITEDSource on GitHub