OBSERVATORY

Signal index.

A cross-domain index of published L2ET research signals, each linked to a practical tool, framework, or source page.

27 published signalsAI Governance: 5Aerospace Engineering: 6Agentic AI: 1Astrophysics: 2Cybersecurity: 3DFIR: 3Enterprise Resilience: 2Frontier AI: 2Frontier and Theoretical Physics: 7Medical AI: 2Post-Quantum Cryptography: 1Quantum Computing: 2RAG Security: 1Robotics: 4Scientific AI: 5Standards and Regulation: 2Theoretical Physics: 4

← back to Observatory

SIGNALS

Published signals.

Use this index when you want the full signal list rather than a domain-specific layer.

evaluation pattern · L2ET Research Method Signal

Frontier AI evaluation must separate capability, autonomy and tool authority

Evaluation plans should test task success, refusal boundaries, tool-use failure, deception pressure, escalation behavior and monitoring before production use.

confidence: high action: apply score: 93

Frontier AI, AI Governance

Signal index.

Published signals.

Frontier AI evaluation must separate capability, autonomy and tool authority

Agentic systems need explicit permission boundaries before deployment

RAG reliability begins with source trust and contamination mapping

AI governance decisions need evidence artifacts, not only policy language

Cyber resilience depends on telemetry coverage before incident response starts

DFIR timelines fail when clocks, evidence sources and log integrity are not controlled

Threat hunting needs falsifiable hypotheses rather than keyword fishing

PQC transition begins with cryptographic inventory and evidence

Quantum computing claims should be checked against qubits, depth and error assumptions

Single-qubit intuition improves when amplitudes are converted into probabilities and Bloch vectors

Effective theories should declare their scale separation and validity domain

Dimensional analysis is a low-cost falsification filter for theoretical models

Speculative physics needs correspondence limits and falsifiable observables

Astrophysical inference must track degeneracy, calibration and selection bias

Observation planning benefits from explicit signal-to-noise and distance assumptions

Orbital design starts with energy, period and transfer margins

Propulsion feasibility should expose mass ratio and delta-v pressure early

Aerospace models should state Reynolds, Mach and Knudsen regimes

Spacecraft communication assumptions need path-loss and margin visibility

Robotics autonomy requires safety cases, fallback and human proximity assumptions

Control behavior should expose damping, overshoot and settling time

Sensor fusion failures often begin with time sync, calibration and covariance drift

Scientific AI credibility depends on reproducibility records

Medical and imaging AI requires evidence discipline before operational claims

Research ideas become durable when ownership, telemetry and recovery are designed

Standards mapping is useful only when it maps to evidence and decisions

Formal methods need stable interfaces, invariants and failure semantics