Coverage map
What complete memory needs to prove.
24 capabilities. 20 modules. Evidence for how they work together.
Module stages
Implementation stage from the audit text. Development code is not an admitted result.
Most connected modules
How many of the 24 capabilities each module helps to test.
Capability matrix
Each row is a memory capability; each column a test module. Violet: the module has development source. Grey: plan only.
Modules and remaining evidence
What exists today and what each module still needs.
Capture/durability
Local development pilot exists.
What it still needs
broader backend and failure/durability evidence remains.
Retrieval/organization
Registered development cell and replay exist.
What it still needs
full external/comparative acceptance remains.
Temporal evolution
Valid-time slice exists.
What it still needs
full transaction-time bitemporality remains.
Conflict/correction
Development matrix and replay exist.
What it still needs
full public acceptance remains.
Provenance/explanation
Development cell exists.
What it still needs
quarantines and complete outcome/lineage evidence remain.
Consolidation/learning
Stage A/B development cell exists.
What it still needs
sustained learning and non-degradation acceptance remains.
Retention/rehearsal/decay
Public CLI rehearsal scheduling tested across five calendar starts spanning over a virtual year.
What it still needs
full seeded retention, retrieval, interference, storage/cost and resource admission remain open.
Reversible forgetting
Public tombstoning exists.
What it still needs
its reversible-delete adapter contract is unresolved. Dedicated fixture, scorer and integration remain missing; restore correctness is not claimed.
Declared-surface erasure
Plan exists.
What it still needs
full benchmark and explicit surface/restore evidence missing.
Calibration/abstention
Deterministic pilot exists.
What it still needs
model-backed calibration and public-label evidence remain.
Security/isolation
Plan and separate security-development infrastructure exist.
What it still needs
full M11 benchmark missing.
Prospective action
Development captures include 130 five-trigger receipts, recovery with 50 injected response losses, and 1,220 bounded fan-out operations with 750 receipts and exact replay. A separate resource run sampled at most 88 MiB RSS.
What it still needs
this is not resource admission. The 220-conversation corpus has a public formation bridge, state/timing diagnostics and saved-trace verification. Local Qwen3 0.6B attempts remain incomplete. The schema variant scheduled no tasks and missed 17 eligible occurrences across 21 completed cases before an HTTP error; this is not a full-corpus score. A single-input follow-up exposed schema-order sensitivity, not correct formation. Draft reference captures remain unadmitted, with a verified nested-condition policy difference. Live PostgreSQL retry/BurnOS checks passed at 55ae2ef9. Full overload/implicit execution, reference calibration, cost and admission remain open.
Working memory
Development probe exists.
What it still needs
capacity and promotion-control experiment remains.
Procedural task utility
Plan exists.
What it still needs
closed-agent environment, policy/baseline/inference contract and implementation remain.
Determinism/replay
Pilot composition exists.
What it still needs
all admitted payloads and real clean-checkout reproduction remain.
Backend/transport parity
Bounded local cassette exists.
What it still needs
broader pairs, migration, resource and acceptance evidence remain.
Custody/recovery
Plan exists.
What it still needs
benchmark implementation and declared fault/recovery scope remain.
Interoperability
Plan exists.
What it still needs
benchmark implementation and fair cross-system interchange contracts remain.
Multimodal memory
Plan exists, module deferred.
What it still needs
portable modality contracts, rights, resources and implementation remain.
Publication integrity
Result-v1/v2, signed ledger, renderer and publication/readiness infrastructure exist.
What it still needs
operational admission and full public evidence remain.
Test the interactions too
Passing each module alone does not prove they work together. These combinations are planned, not measured.