smrti

# Salience-ranked retrieval
results = engine.recall("deployment risks")

for r in results:
    print(
        f"{r.atom.content}",
        f"salience={r.salience:.2f}",
        f"confidence={r.atom.truth.confidence:.2f}",
        f"valence={r.atom.valence.valence:.1f}",
    )

# KNN + BM25, fused → 1-hop graph expansion → salience → diversity cap
# Negative-valence atoms surface first when relevant

# Assert a belief with supporting evidence
engine.believe(
    "Feature flags reduce deployment risk",
    probability=0.85,
    evidence="3 successful staged rollouts"
)

# Evidence is appended, never overwritten;
# engine.evidence(atom_id) reads the log back.
# The consolidation epoch revises truth via PLN.

# Report that recalled memories were used
results = engine.recall("deployment risks")
engine.reinforce([r.atom.id for r in results])

# Use is weak evidence: confidence climbs a little,
# capped per epoch. Probability does not move.

# Run a consolidation epoch
epoch = engine.reflect()

print(epoch)
# EpochResult(
#   beliefs_updated=3,
#   atoms_decayed=12,
#   atoms_pruned=1,
#   lti_promoted=2,
#   new_connections=4,
#   contradictions_resolved=1,
#   orphans_healed=2
# )

# Revise evidence → decay → propagate → heal → promote
# → resolve contradictions → prune. Runs when the space is used.

# Stop matching memories from surfacing
forgotten = engine.forget("Friday deployment outage", top_k=5)

# Confidence sinks below the surfacing floor and the atoms
# are stamped: excluded from every recall, never lifted back
# by consolidation, and removed by a later epoch.

# Switch preset at runtime
engine.set_personality("analytical")

# 6 presets: balanced, analytical, curious,
# empathetic, maverick, deterministic
# Each tunes 17 hyperparameters that control
# learning rate, decay, promotion, and salience weights