external-belief-ingestion-is-defensively-layered
OUT derived (depth 2)
External beliefs enter the system through defensively-layered pipelines regardless of source: LLM derivation applies fail-soft validation, Jaccard retraction guards, and environment isolation, while agent import provides dual reconciliation modes with heterogeneous truth state handling — both converge on the same underlying mutation infrastructure.
Summary
When new information comes into the system from outside — whether generated by an LLM or imported from another agent — it goes through protective checks before being accepted. LLM-generated beliefs get validated and filtered to avoid bad data, while imported beliefs can either be added alongside existing ones or used to overwrite them. Both paths ultimately use the same core machinery to make changes, meaning the system has a consistent safety layer no matter where new beliefs originate.
Justifications
SL — depth-2 — two independent ingestion paths (LLM derive, agent import) each have their own defense mechanisms, providing defense in depth for all external input
Antecedents (all must be IN):
- derive-pipeline-is-defensive — The derive pipeline applies multiple defensive measures: fail-soft validation, Jaccard-based retraction guard, and environment variable stripping to prevent recursive spawning
- import-sync-has-dual-reconciliation-modes — The import/sync subsystem offers two distinct reconciliation strategies: import is additive (skips existing nodes), while sync is remote-wins (overwrites text, justifications, and truth values from the remote source).
Dependents
These beliefs depend on this one:
- external-beliefs-defensively-contained — External beliefs pass through two independent safety layers: defensive ingestion pipelines (fail-soft validation, Jaccard guards, dual import/sync reconciliation modes) filter and validate beliefs on entry, while the self-contained agent subsystem (namespace isolation, relay-pair kill-switches, reversible lifecycle management) constrains their operational footprint after ingestion.
- external-ingestion-is-format-resilient-and-defensively-layered — External belief ingestion achieves both defensive containment (fail-soft validation, Jaccard retraction guards, dual import/sync reconciliation, namespace isolation) AND format resilience (parser version fallback, forward-compatible metadata parsing, prose-tolerant JSON extraction), ensuring robust integration even as external source formats evolve unpredictably.