onnxruntime-node@1.24.3 exposes network, filesystem, process, crypto, obfuscation, dataExfiltration with overall critical risk.
Overall riskCRITICAL (85)Package-level aggregate score
Source matchLOWConfidence in decompiled/source alignment
Binary inventory14162593 KB total analyzed size
Behavior familiesnone detectedObserved across all binaries
Risk score85critical severity posture
Binary count14Native artifacts recovered
ConfidenceLOWSource/decompile alignment
Total size162593 KBCombined binary payload
Analyst takeaways
3 evidence summariesNo escalated findingsThe current evidence set is dominated by expected native behavior.
0/14 binaries carry findingsNot every native artifact in a package deserves equal attention. Focus review where findings cluster.
binshield-worker analysis with low confidenceUse the model summary as triage guidance, then validate against imports, strings, and recovered functions.
Version history
1 analyzed versions- Confidence: low
- Analysis model: binshield-worker
- Data mode: Connected to API
Version drift
5.1.6 to 5.1.7This version introduces additional behavior that deserves security review.
Version 5.1.7 adds stricter extension loading checks and a slightly larger native payload.
Risk increased by 4 pointsUse this as triage guidance, then validate against binary evidence below.
Added: Additional filesystem path validation before extension loading.Added: Guarded extension loading before execution.
- Validate whether any newly added filesystem or process behavior is expected.
- Compare binary inventory to ensure no unexpected native artifact was introduced.
- Review high-signal strings and imports before approving rollout.
Finding clusters
0 findingsNo escalated findings were emitted for this package version.
Binary evidence
14 native artifactsRelated packages
Similar signal surfaceNo related packages surfaced from the current dataset.
How this score was computed
Methodology transparencynone0-0
low1-29
medium30-59
high60-79
critical80-100
85
NONE0-0No behaviors or findings detected.
LOW1-29Expected behaviors only. Safe for most environments.
MEDIUM30-59Some review-worthy behaviors. Inspect before deploying in hardened environments.
HIGH60-79Multiple risk signals. Manual review required before production use.
CRITICAL80-100Severe risk indicators. Block until validated by a security engineer.
Scoring factors
Findings2-45 pts per findingSeverity-weighted: info=2, low=8, medium=18, high=30, critical=45
Behaviors3-28 pts per detectednetwork=14, filesystem=4, process=12, crypto=3, obfuscation=24, exfiltration=28
Import countUp to 6 ptsimportCount / 4, capped at 6. More imports = larger attack surface.
Function countUp to 5 ptsfunctionCount / 20, capped at 5. Complexity indicator.
Package-level score = 65% highest binary score + 35% average binary score. Scores are deterministic and reproducible.