The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes pathogen data and nomenclature retrieved from external API instances and public GitHub repositories. While this introduces an attack surface where a hostile data source could potentially provide malicious instructions, the skill implements a sanitize function in scripts/lapis_client.py that strips C0 and C1 control characters and collapses whitespace. This prevents remote data from being interpreted by the agent as structural commands or forged output.
Ingestion points: Pathogen genomic data from GenSpectrum LAPIS API instances (e.g., lapis.cov-spectrum.org) and the cov-lineages/pango-designation repository on GitHub.
Boundary markers: Results are presented to the agent in structured formats (Table, TSV, JSON) with clear provenance headers containing data versions and filters.
Capability inventory: The agent environment allows the use of Bash, Read, Write, and Edit tools.
Sanitization: The sanitize() utility is applied to all remote strings to filter non-printable control characters and prevent newline injection.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to retrieve live genomic surveillance data from GenSpectrum and Pathoplexus LAPIS instances, as well as nomenclature mappings from the Pango Designation project's GitHub repository. These destinations are well-known scientific resources necessary for the skill's primary function of providing up-to-date pathogen information.

pathogen-variant-surveillance