vibe-coding-db-duckdb
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for ingesting data from untrusted external sources, such as CSV, Parquet, and JSON files, which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: Data is read from local file paths and remote URLs using functions like
read_csv,read_parquet, andread_jsoninSKILL.md. - Boundary markers: The code templates do not include specific delimiters or instructions to prevent the agent from following commands embedded within the data files.
- Capability inventory: The skill demonstrates file-writing capabilities using the
COPY TOSQL command inSKILL.md. - Sanitization: No sanitization or validation of the external file content is implemented before processing.
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install standard libraries and demonstrates fetching data from external URLs.
- Fetches packages from the official NPM registry:
duckdb,@duckdb/node-api. - Fetches packages from the official PyPI registry:
duckdb. - Demonstrates querying remote data from an external HTTP source (
https://example.com/data.parquet).
Audit Metadata