Corpus Behavioral¶
Overview¶
The package's position in the corpus, or its deviation from a corpus
baseline. H046, H057, H060, H061, and H073 are Class D rules that run once
per metadata cycle in trustsight full-aur.
They are silent without a prior snapshot, which is enforced rather than
assumed: the calibration gate is fire_rate(no_baseline) == 0. H061 and
H057/H060 are prioritisation rather than accusation; they say what a
compromise would reach, not that anything is wrong.
See Rule Mechanics for the field table, the severity weights and the reserved identifier ranges.
Threat context¶
You know a friend has been replaced by an impostor not because they did anything obviously wrong, but because they do not sound like themselves. A teacher catches a ghostwritten essay the same way: the voice is subtly off. The corpus-behavioral rules work on that instinct. Most rules ask "is this pattern dangerous?"; these ask "does this package behave like the thousands of benign packages we have seen?" The model is trained on the benign corpus, so what it flags is novelty - the package that does not look like its peers, whether or not any human named the pattern.
This is the learned layer of TrustSight, and its limits are the corpus's limits: it can only know normal from the examples in the priors, so a rare-but-legitimate build can read as novel. It is deliberately a mild-signal contributor that raises suspicion for named rules to corroborate.
Detection mechanisms¶
The rules that make up this category are listed below, each with the series that implements it and the severity it carries. Scoring, matching and identifier allocation are shared across every category: see Rule Mechanics.
Individual rules¶
| Rule | Name | Severity |
|---|---|---|
| H028 | Accelerated Release Cadence | - |
| H046 | Orphan/Adoption Dependency | MEDIUM |
| H057 | Transitive Exposure | INFO |
| H060 | Transitive Orphan Risk | INFO |
| H061 | Dependency Centrality | INFO |
| H073 | Introduction Rate Deviation | MEDIUM |
H028: Accelerated Release Cadence¶
H028 is metadata, never a scored finding. The pipeline records whether the
HEAD commit has 3+ commits within 24 hours (counting HEAD itself) as the
recent_commit_burst boolean on PackageFact; it is not appended to
triggered_rules and contributes nothing to the score.
H046: Orphan/Adoption Dependency¶
- Severity: MEDIUM (weight 15)
-
Category:
dependency -
Condition: A package depends directly on a package orphaned or adopted in this cycle.
H057: Transitive Exposure¶
- Severity: INFO (weight 0)
-
Category:
dependency -
Condition: A package's transitive closure, at least
[thresholds] h057.min_hops(default 2) hops away, reaches a package adopted out of the orphan state this cycle. Context only, never additive.
H060: Transitive Orphan Risk¶
- Severity: INFO (weight 0)
-
Category:
dependency -
Condition: A package's transitive closure includes a currently orphaned package. Context only, never additive.
H061: Dependency Centrality¶
- Severity: INFO (weight 0)
-
Category:
dependency -
Condition: A package is depended on by at least
[thresholds] h061.min_dependents(default 50) AUR packages. Prioritisation only: it says what a compromise would reach, not that anything is wrong.
H073: Introduction Rate Deviation¶
- Severity: MEDIUM (weight 15)
-
Category:
adoption -
Condition: The corpus-wide introduction rate for a cycle deviates from the baseline mean by at least
[thresholds] h073.z_score(default 3.0), once at leasth073.min_history_cycles(default 3) cycles of history exist. An immature history is quiet.
Interpretation¶
- The finding says the package deviates from the benign corpus's behaviour, with the feature(s) driving it. It is not a named attack.
- A lone novelty finding is weak; novelty plus a concrete rule is the case where the corpus adds real value.
- If the priors were trained on a narrow slice, novelty findings will be noisy; this is a corpus-provenance question, not a package one.