The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides standard command patterns for workflow execution and environment verification. Examples include nextflow run main.nf and generic Python version checks (python -c "import <module>; print(<module>.__version__)"). These are routine for the skill's stated purpose of pipeline orchestration.
[INDIRECT_PROMPT_INJECTION]: The skill has an inherent attack surface because it is designed to ingest and process untrusted external data such as pipeline definitions (main.nf) and sample sheets (samplesheet.csv).
Ingestion points: External file inputs referenced in SKILL.md (samplesheet.csv, main.nf, environment descriptions).
Boundary markers: The skill instructs to 'State inputs, outputs, parameters, and expected execution environment clearly', but does not specify strict delimiters for the data content itself.
Capability inventory: The skill uses subprocess execution via the nextflow CLI and python (found in SKILL.md).
Sanitization: The skill includes defensive instructions such as 'Verify config, manifests, and sample sheets before launching a full run' and 'Confirm identifiers and metadata join correctly'. This mitigates the risk of processing malformed or malicious metadata.

workflow-management