Directories
¶
| Path | Synopsis |
|---|---|
|
Command analyse derives evaluation results from a replay run's result JSON.
|
Command analyse derives evaluation results from a replay run's result JSON. |
|
Command budgetcurve turns a framework replay and a baselines run into §1.1's figure: one line per detector through its own operating points, across the whole budget range.
|
Command budgetcurve turns a framework replay and a baselines run into §1.1's figure: one line per detector through its own operating points, across the whole budget range. |
|
Command corpuscheck verifies the derived corpus against the digests the recorded runs cite, and fails closed.
|
Command corpuscheck verifies the derived corpus against the digests the recorded runs cite, and fails closed. |
|
Command dashboard renders an interactive evaluation dashboard from result JSON.
|
Command dashboard renders an interactive evaluation dashboard from result JSON. |
|
Command e5 runs the §11.3 schema-growth experiment (hypothesis E5).
|
Command e5 runs the §11.3 schema-growth experiment (hypothesis E5). |
|
Command e7 measures evaluation hypothesis E7 (§12.3): the proportion of sampled verdicts reconstructable from their evidence alone, with no query back to the store.
|
Command e7 measures evaluation hypothesis E7 (§12.3): the proportion of sampled verdicts reconstructable from their evidence alone, with no query back to the store. |
|
Command e8 executes evaluation hypothesis E8 (§12.3) as a recorded run and writes its result JSON:
|
Command e8 executes evaluation hypothesis E8 (§12.3) as a recorded run and writes its result JSON: |
|
Command features exports the numeric feature vectors the baseline detectors of §12.4 consume: Isolation Forest, Extended Isolation Forest, Half-Space Trees, and Robust Random Cut Forest, run by sidecar/baselines.py.
|
Command features exports the numeric feature vectors the baseline detectors of §12.4 consume: Isolation Forest, Extended Isolation Forest, Half-Space Trees, and Robust Random Cut Forest, run by sidecar/baselines.py. |
|
Command inject plants synthetic attacks of named kinds into a corpus, so detection can be measured per attack type against controlled ground truth.
|
Command inject plants synthetic attacks of named kinds into a corpus, so detection can be measured per attack type against controlled ground truth. |
|
Command methodtable builds the paper's headline comparison: one row per method, one column per attack type.
|
Command methodtable builds the paper's headline comparison: one row per method, one column per attack type. |
|
Command replay streams a corpus through the framework and writes a result JSON carrying full provenance: run identity, git state, corpus checksums, row counts, the frozen burn-in split, parameters, and the measured outcomes.
|
Command replay streams a corpus through the framework and writes a result JSON carrying full provenance: run identity, git state, corpus checksums, row counts, the frozen burn-in split, parameters, and the measured outcomes. |
|
Command runreport produces a per-run report folder from a single experiment result JSON file.
|
Command runreport produces a per-run report folder from a single experiment result JSON file. |
|
Command subset writes a smaller corpus in the source format, so that a full pass of the framework can be rehearsed in minutes rather than hours.
|
Command subset writes a smaller corpus in the source format, so that a full pass of the framework can be rehearsed in minutes rather than hours. |
|
Command thesis renders docs/PAPER.md as a self-contained HTML document with diagrams.
|
Command thesis renders docs/PAPER.md as a self-contained HTML document with diagrams. |
|
Command volumegate turns the volume_gate_probe block of an ungated replay into the committed decision record for #25: which completed-period threshold the volume arm's R3 abstention is set to, and what every rejected candidate would have done instead.
|
Command volumegate turns the volume_gate_probe block of an ungated replay into the committed decision record for #25: which completed-period threshold the volume arm's R3 abstention is set to, and what every rejected candidate would have done instead. |
|
Command wraparound executes the §12.5 wraparound control as a recorded run and writes its result JSON, including the fitted curves that figures F1 and F8 render.
|
Command wraparound executes the §12.5 wraparound control as a recorded run and writes its result JSON, including the fitted curves that figures F1 and F8 render. |
Click to show internal directories.
Click to hide internal directories.