Bound-gate test fixture CANDIDATE
A separate synthetic experiment at bc0ee13. Its recorded Python result is EXECUTE under a declared test root. It does not change the original source fixture’s HOLD.
The first tables show a recorded Python verifier run. A separate browser evaluator below recomputes its checks. Neither governs an external effect, establishes production authority, or replays a historical source decision. The signer’s seed is public. DECISION_REPLAY_UNAVAILABLE.
File manifest · Recorded check report (JSON) · Independent Python verifier · Experiment README
Recorded test-domain positive
EXECUTE bound-gate-valid. All eight reported checks PASS; the warrant applies only under this experiment’s test root.
| Check | Recorded status | Meaning |
|---|---|---|
byte_conformance | PASS | strict decode, canonical re-encode, hashes and published fields match |
signature | PASS | Ed25519 verifies under key ff7334e9.. over domain||verdict |
input_binding | PASS | verdict commits to exactly this request+context |
material_availability | PASS | content hashes of retrievable policy & authority-state match commitments |
definition_binding | PASS | published definitions equal the verifier's registered test predicates |
invariant_evaluation | PASS | measured mask 0x07 vs claimed 0x07; required ['BG-I1', 'BG-I2', 'BG-I3'] -> all pass; outcome supported; required set registered |
measurement_binding | PASS | published measurement agrees with recomputed results and mask |
test_authority | PASS | TEST authority: key endorsed by declared non-illustrative trust root 'bound-gate experimental test trust root' and by authority-state (production_identity_claimed=False) |
Five counterexamples
| Mutation | Result | Failing check |
|---|---|---|
input-binding-mismatch | HOLD | input_binding |
policy-bytes-tampered | HOLD | material_availability |
authority-state-tampered | HOLD | material_availability |
invariant-measurement-mismatch | HOLD | invariant_evaluation |
signature-invalid | HOLD | signature |
Fresh browser evaluation
The browser retrieves and hashes the published files, then runs a separate TypeScript evaluator. Its result concerns the experiment’s declared test authority only. It is not a replay of the historical source gate, and a site serving consistently replaced files could also replace the manifest.
Fetching and checking files…
Inputs and provenance
Source: crossing-graph-design@bc0ee13a72734cdad639cd428d6cb3360f16fc6a, path experiments/bound-gate-fixture. The signed verdict binds to a request/context pair and content-hashed policy and authority-state files. The definitions and published measurements are compared to independently recomputed test predicates, but those readable metadata fields are not themselves signed by the source GateVerdict.
| File | SHA-256 |
|---|---|
| README.md | 67ff5da0725ad65b0eae9a42d3980dd8c18abb6bdc7ccf2df9350541abb5f164 |
| bound-gate.v1.json | 7e74cae18bae6c26f9f64457da89461d761f8ac7005c849dd08d141261275daf |
| generate.py | cbf82551431747e9c2c8ed02d3dcb0ed06e3524effedde0af1ff2ccbebeaa8f1 |
| material/authority-state.json | fdaa4d5a01de28341844a0b24d8255347fb33a92323794e00279db03f9203bfb |
| material/policy.json | ff2723aa07000d29ebf096c95f631e13f757aed318faa62c0873187ae4c8b8c8 |
| report.v1.json | 62f3cd3ba5dc7ea4a74ddccc21da22b74f8719cd37dcbad862c6017bcdbb9664 |
| requirements.txt | e694c9f1e9be3eab5524d866c26eb518df953b371ea708a174e200f22e623c07 |
| test_bound_gate.py | 472fc05d245f2e1db28eeeb4a5aec1174eedb3f46c4b68f70e013ce0956e839b |
| trust-root.json | 8c628119bc07d007a72be72add5cb84225ff7bc7a46bd7802c3ce72cb3409c05 |
| verify_bound_gate.py | 1858966b2aa286aca3ae1adeb4ad2ecf2c6bf6d1455a4fd0dec302ad319e4ed4 |
Manifest and files share one site origin. Their hash matches detect accidental drift, not a host serving a coordinated replacement. In a local checkout with Python and cryptography, run python public/experiments/bound-gate-test-001/verify_bound_gate.py for a separate local test run.