The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of VCF (Variant Call Format) files which are untrusted external data sources. Maliciously crafted VCF headers or data records could contain instructions intended to override or influence the agent's behavior.
Ingestion points: The skill ingest data from input.vcf using commands like bcftools view -h and bcftools query.
Boundary markers: There are no instructions to use delimiters or ignore embedded instructions when reading or processing VCF content.
Capability inventory: The skill utilizes tools such as bcftools, awk, and grep which possess file reading and data manipulation capabilities.
Sanitization: The instructions do not include steps for sanitizing or validating the content of the VCF file for potential prompt injection payloads before processing.

vcf-variant-filtering