ANMA-X v1.0
NORMAL
RUNNING t = 0
ANMA-C · Cognitive
αtExploration drive0.62
βtExploitation / focus0.71
δtUncertainty sensitivity0.34
Epistemic state
γtConfidence0.78
Cross-domain coupling
κ ← κ₀ + k·(1−γ)
0.30 + 0.40·(1−0.78)
κ_eff = 0.39
High-risk candidates
trigger δ escalation →
additional eval cycles.
LLM / Cognitive Agent reasoning · planning · attention c_t, s_t ANMA-X · Regulatory Middleware ANMA-C α · β · δ cognitive mod. γ=0.78 → δ↑ ⊕ ANMA-A κ · λ · φ · ψ action mod. π(a|s) inhibition π(a|s) ✓ Actuator Layer tool use · file ops · APIs · env feedback π(a|s) = softmax( f(c,s) − g(a,μ) )
Execution log 0 entries
ANMA-A · Action Queue
Neuromodulators
κtIrreversibility gate0.39
λtExternal-risk sensitivity0.45
φtPlan-chain damping0.28
ψtConsensus weighting0.60
Safety properties
◉ Monotonic risk inhibition
◉ Confidence-dep. threshold
◉ Consensus-sensitive actuation
◉ Plan-length regularization
μ = (α,β,δ,κ,λ,φ,ψ) | g(a,μ) = – | actions pending: 0