The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions to download and execute installation scripts directly from official remote sources for the Nextflow engine and the nf-test framework.
Evidence: curl -s https://get.nextflow.io | bash in SKILL.md.
Evidence: curl -fsSL https://get.nf-test.com | bash in references/testing.md.
[COMMAND_EXECUTION]: The skill instructs the user to use elevated privileges to move the Nextflow executable to a system-wide path.
Evidence: sudo mv nextflow /usr/local/bin/ in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of official tooling via standard package managers and direct downloads from well-known technology domains.
Evidence: Fetches nf-core via pip and conda as described in SKILL.md.
[PROMPT_INJECTION]: The skill describes patterns for ingesting external data, such as CSV samplesheets, which represents an inherent surface for indirect prompt injection common to workflow management tools.
Ingestion points: Data is ingested from external files using channel.fromPath and splitCsv as described in references/language.md.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content embedded in the processed data files.
Capability inventory: The Nextflow environment allows for the execution of arbitrary Bash commands and network access within process blocks.
Sanitization: The documentation does not specify input validation or sanitization steps for external data before it is interpolated into shell commands.

nextflow