Dimensional Drift Witness is organized as a deterministic proof artifact.
The architecture separates case data, governing policy, verifier logic, generated receipts, proof-pack validation, CLI access, and CI replay.
drift_cases.jsonsupplies paired execution states.drift_policy.jsondefines the severity model.drift_verifier.pycompares operational continuity against legitimacy continuity.drift_receipt.jsonrecords the deterministic witness result.proof_pack.pyverifies expected proof counts and receipt structure.tests/validates discrimination behavior..github/workflows/proof-pack.ymlruns proof replay in CI.
Operational dimensions:
- actor
- action
- output
- replay trace
Legitimacy dimensions:
- authority
- custody
- standing
- scope
The witness only classifies dimensional drift when operational continuity holds while one or more legitimacy dimensions diverge.
DRIFT_DETECTED: operational continuity holds and legitimacy continuity fails.NO_DRIFT: operational and legitimacy continuity both hold.OPERATIONAL_DIVERGENCE: operational continuity fails, so legitimacy drift cannot be isolated.
The repository does not enforce production decisions directly.
It produces a receipt indicating what enforcement response would be required if the witness were integrated into a production authority and custody system.
Running the proof pack must not mutate the repository state.
Expected replay check:
python proof_pack.py
python -m pytest -q
git status
Expected result:
nothing to commit, working tree clean