cgh-classify

Frozen. 0.2.0 is the final release. Its main consumers, the core egress gate's confidentiality labels and the guard, went away with secure mode in cgh 0.15.0, and cgh[plugins] no longer installs it. It still works and still installs by name (pip install cgh-classify), gets no new features, and may be removed from the cgh repository later. cgh-codegen's egress gate still honors a confidential finding when one exists.

Human-trainable confidentiality classification for cgh. You label a few files, a lightweight local model (TF-IDF + naive Bayes, standard library only) generalizes to the rest, and the result lands as confidential findings. Nothing ever leaves the machine.

pip install cgh-classify
cgh classify label payroll.xlsx            # mark confidential
cgh classify label README.md --not         # mark public
cgh classify train                         # fit + sweep the repo
cgh classify review                        # files the model is unsure about
cgh findings --key confidential

How labels and predictions interact

The asymmetry is deliberate: a model may block on its own say-so (worst case, a false positive costs a summary), but only a human label can clear a file under a strict egress gate (egress = "strict"), where the gate is an allowlist.

Configuration

cgh classify label and cgh classify train write findings when you run them. Classifying every file on each index is opt-in since 0.2.0:

[plugin.classify]
# scan_on_index = false  # re-classify every indexed file
# threshold = 0.7      # predict confidential above this probability
# uncertain_low = 0.35 # review window lower bound
# uncertain_high = 0.65

Your labels are the asset: they live in .codegraph/classify_labels.json, the trained model in .codegraph/classify_model.json, both machine-local and cheap to retrain (cgh classify train is instant on thousands of files).