ecological-data-foundation
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill performs legitimate data retrieval from well-known biodiversity repositories including GBIF (api.gbif.org), iNaturalist (api.inaturalist.org), OBIS (api.obis.org), and the IUCN Red List (apiv3.iucnredlist.org). These operations use official APIs and specialized libraries designed for these services.
- [DATA_EXPOSURE]: Several scripts (e.g., download_from_gbif.py, download_from_iucn.py) access environment variables for API credentials (GBIF_USER, GBIF_PWD, GBIF_EMAIL, IUCN_REDLIST_KEY). This is documented as a requirement for accessing protected data services and follows standard secure coding practices for secret management.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to ingest and process external datasets (CSV, XLSX, TSV, EBD). A malicious dataset could theoretically contain hidden instructions intended to influence the agent when it summarizes the cleaning reports or quality assessments.
- Ingestion points: Data ingestion occurs in scripts like
clean_occurrences.pyanddownload_from_ebird.pywhich read local files into dataframes. - Boundary markers: The skill produces structured outputs (CSV) and summary reports (Markdown), but does not explicitly use defensive delimiters for the agent context.
- Capability inventory: The skill possesses file write capabilities and network access to official biodiversity APIs.
- Sanitization: The skill relies on standard data parsing libraries (pandas, readr) for validation, which is typical for the intended use-case of processing scientific data.
Audit Metadata