lamindb
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external biological data files, creating an indirect prompt injection surface.\n
- Ingestion points: Data enters the agent's context through functions such as ad.read_h5ad(), ln.Artifact.get(), and artifact.load() in SKILL.md.\n
- Boundary markers: The provided code examples do not include explicit instructions or markers to disregard embedded commands within processed data.\n
- Capability inventory: The skill allows for filesystem operations, interaction with cloud storage (S3/GCP), and execution of workflow managers like Nextflow.\n
- Sanitization: The skill recommends the use of AnnDataCurator and biological ontology standardization for data validation, providing basic integrity checks.\n- [SAFE]: The skill uses legitimate Python packages and references well-known official documentation.\n
- External Links: References to lamin.ai, official GitHub repositories, and k-dense.ai are appropriate for the skill's purpose.\n
- Dependencies: Software dependencies such as lamindb, bionty, anndata, and wandb are standard libraries in the bioinformatics and research community.
Audit Metadata