The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process large-scale datasets from various external formats, creating a surface for indirect prompt injection if the data contains malicious instructions.
Ingestion points: The skill uses dd.read_csv, dd.read_parquet, and db.read_text to load data from local and remote storage (SKILL.md, references/dataframes.md, references/bags.md).
Boundary markers: The documentation does not provide specific instructions for using delimiters or boundary markers to isolate data from agent instructions.
Capability inventory: The skill utilizes Bash for installation and execution, along with network-enabled cloud storage drivers (s3fs, gcsfs) and file-writing capabilities (to_parquet, to_csv).
Sanitization: No explicit data sanitization methods for preventing prompt injection are mentioned.
[DYNAMIC_EXECUTION]: Dask's architecture is fundamentally based on distributing and executing Python code across worker nodes, which is a form of dynamic execution.
Evidence: The skill extensively documents the use of client.submit, client.map, and collection-level map functions (map_partitions, map_blocks) to execute arbitrary Python functions (references/futures.md, SKILL.md).
[COMMAND_EXECUTION]: The skill provides shell commands for environment management and package installation.
Evidence: uv pip install commands for Dask and its dependencies are included in the Quick Start section of SKILL.md.

dask