scvelo
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill demonstrates standard bioinformatics workflows for RNA velocity analysis. No malicious code, obfuscation, privilege escalation, or persistence mechanisms were identified. The logic is consistent with the stated purpose of analyzing single-cell RNA sequencing data.
- [EXTERNAL_DOWNLOADS]: The skill references the
scvelopackage and official documentation hosted on GitHub and ReadTheDocs. These are well-known, reputable resources in the bioinformatics community and are required for the skill's functionality. - [INDIRECT_PROMPT_INJECTION]: The skill includes functionality to ingest external scientific data files (Loom and H5AD).
- Ingestion points: Biological data is loaded using
scv.read()andsc.read_h5ad()inSKILL.mdandscripts/rna_velocity_workflow.py. - Boundary markers: Not applicable, as these files are structured binary/scientific data formats rather than natural language text.
- Capability inventory: The skill's capabilities are restricted to local file operations (reading data and saving analysis results/plots) and standard scientific computation. It lacks network exfiltration or shell command execution capabilities.
- Sanitization: The ingested data is processed as numerical matrices and scientific metadata, which does not provide a path for natural language instruction override.
Audit Metadata