Provenance chains

Every memory carries derived_from and supersedes edges to its ancestors. memledger walks the chain on search() to compute weakest-link effective confidence and surfaces the chain in the audit trail.

The chain is what enables memledger's most opinionated guarantee: a high-confidence claim built on a low-confidence ancestor cannot outscore its weakest link.

# Triage agent makes a hedged guess
triage_id = await ml.add(
    content="possible connection-pool exhaustion",
    confidence=0.30,
    hedged=True,
    created_by="triage-agent",
)

# Diagnosis agent confidently asserts a fix derived from triage
await ml.add(
    content="bump maxPoolSize from 10 to 50",
    confidence=0.85,
    created_by="diagnosis-agent",
    derived_from=[triage_id],   # this is the chain
)

When the diagnosis record is later retrieved, its effective confidence is min(0.85, 0.30) = 0.30 — the chain is honest about the weakest upstream link.

For the conceptual framing — including the DAG semantics and the JsonArrayChainStore vs Neo4jChainStore trade-off — see Provenance chain.

The full configuration reference — max_hops tuning, ChainStore Protocol implementation — ships in a future release.