Workspace settings & snapshots
Configuration & prerequisites
Immutable snapshot history
Snapshot provenance / analysis coverage
No analysis recorded.
Follow the data.
Explore connections. Keep what matters.
No selected types means all types. Use Ctrl/Command to select multiple options. The explicit node limit applies after filtering; only edges between displayed nodes are drawn. Sources, sinks, functions and calls appear first.
Knowledge graph
Analysis results & policy
Analysis does not repair code. Runtime results cover executed synthetic cases only. Model advice cannot confirm a finding, approve policy, or pass a repair gate. Repairs are isolated local branches and local merge-request simulations, never a push, merge, or deployment.
Executed runtime evidence
Sanitized runtime record
No runtime evidence recorded.
Luna boundary decisions
Connected source and graph evidence, including boundaries with no static finding. Probabilities are not leak confirmation or safety approval; missing context remains unresolved.
Complete review contexts and typed model responses
No Luna decision recorded.
Run metrics & labeled evaluation
Recent run metric records
No runs recorded.
Measured labeled-case evaluation
Precision/recall are measured only against configured labeled cases, not inferred by the model or from an unlabeled repository.
No labeled evaluation recorded.
Independent repair gate
No repair recorded.
Independent verification details
No verification recorded.
Proposed / applied isolated diff
No diff recorded.
Policy lifecycle & explicit operator approval
Policy approval is a human action, separate from model advice and repair verification. Review the exact content before approving. Stored policy versions are immutable; approval does not silently rewrite an existing version or activate a different configured policy file.
No policy loaded.