The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from external sources such as S3 buckets and log files, which represents a potential surface where processed data could contain instructions. However, this functionality is fundamental to the library's purpose for large-scale data analysis.
Ingestion points: Data loading via dd.read_csv('s3://my-bucket/data.csv') and db.read_text('logs/*.log') as documented in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or warnings for the agent to ignore instructions embedded within external datasets.
Capability inventory: The skill utilizes task scheduling and custom function parallelization through dask.delayed and client.compute to execute computation graphs.
Sanitization: No data sanitization or validation logic is implemented for the external inputs within the provided examples.
[EXTERNAL_DOWNLOADS]: The skill provides standard instructions to install the dask[complete] package from the official Python registry via pip. This is a legitimate installation step for a well-known library.
[COMMAND_EXECUTION]: The documentation includes code snippets for initializing a local Dask cluster and executing distributed tasks, which are standard operations for development and data processing with this library.

dask