onekgpd
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the
dnaerysPython package (v0.2.1) and communicates with a remote gRPC service atdb.dnaerys.org:443to retrieve genomic and variant data. - [COMMAND_EXECUTION]: The skill utilizes
uv runto execute its core Python scripts (onekgpd_api.pyandonekgpd_meta.py), which triggers automatic dependency resolution and execution of local code. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, which could potentially contain malicious instructions targeting the LLM.
- Ingestion points: External genotypes and variant annotations are streamed from
db.dnaerys.org:443viaonekgpd_api.py. Sample metadata is loaded from the localassets/kgpe.jsonfile viaonekgpd_meta.py. - Boundary markers: The tool outputs data into structured JSON files and recommends using
jqfor field extraction, providing a degree of separation between raw data and the agent's prompt context. - Capability inventory: The skill has permissions to write Bash commands and execute Python scripts, which are used to process and summarize the fetched data.
- Sanitization: Results are returned in a structured JSON format; however, the skill does not explicitly sanitize the text content of genomic annotations or metadata fields against prompt injection markers.
Audit Metadata