The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from global news outlets and TV transcripts via the GDELT Project.
Ingestion points: Data is retrieved from https://api.crawlora.net endpoints using the scripts/crawlora.sh helper.
Boundary markers: The skill returns content in normalized JSON format, which provides structure but does not include explicit instructions to the LLM to ignore embedded instructions within the news text.
Capability inventory: The skill has the capability to perform network requests (curl) and local script execution to retrieve and format news data.
Sanitization: User-provided arguments are URL-encoded via curl --data-urlencode in the wrapper script to prevent command or parameter injection into the API request.

gdelt-research