The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates the processing of external, untrusted bioinformatics data files through its workflow system. This represents a potential surface for indirect prompt injection where malformed data could attempt to influence the behavior of the analysis tools or the agent's interpretation.
Ingestion points: Parameters defined as LatchFile or LatchDir in workflow definitions, such as my_workflow(input_file: LatchFile) in SKILL.md.
Boundary markers: The instructions do not provide specific boundary markers or warnings to the agent to disregard potential instructions embedded within the processed genomic data.
Capability inventory: The skill uses @task and @workflow decorators to define and execute serverless code, and supports the registration of Nextflow and Snakemake pipelines.
Sanitization: There is no documentation regarding the sanitization or validation of the internal contents of the genomic files (e.g., FASTQ, BAM) before processing.
[EXTERNAL_DOWNLOADS]: The skill guides the user to install the latch package from the Python Package Index (PyPI). This is the official SDK for the Latch platform described in the skill.

latchbio-integration