The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes legal opinions and docket data from external sources. While this is public legal information, it is untrusted external content that could contain embedded instructions.
Ingestion points: scripts/bulk_fetch.py downloads data from com-courtlistener-storage.s3-us-west-2.amazonaws.com.
Boundary markers: The script does not use explicit delimiters to isolate ingested data from agent instructions.
Capability inventory: The script writes to the local filesystem (--out flag) but does not have execution capabilities or network-send functions for the ingested data.
Sanitization: Standard CSV parsing is performed, but there is no specific sanitization for prompt injection patterns within the opinion full text.
[EXTERNAL_DOWNLOADS]: The skill downloads large compressed CSV datasets from Free Law Project's official AWS S3 bucket. This is a well-known service for legal research data and the downloads are for their intended purpose.
[COMMAND_EXECUTION]: The skill provides documented examples of using the AWS CLI (aws s3 ls) and its own Python script to interact with the public data repository.

courtlistener