The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches investigative data and public records from authoritative organizational sources, including the SEC (sec.gov), U.S. Treasury (treasury.gov), USAspending (usaspending.gov), NYC Open Data (cityofnewyork.us), GDELT Project (gdeltproject.org), ICIJ (icij.org), and Wikimedia (wikipedia.org). These operations are essential to the research purpose of the skill and utilize official or well-known service domains.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the public web (e.g., news article snippets, company descriptions, and litigation summaries) which is subsequently processed by the agent.
Ingestion points: External data is retrieved via scripts like fetch_gdelt.py and fetch_wikipedia.py and stored in CSV format.
Boundary markers: There are no specific delimiters used in the instructions to isolate external content from agent instructions.
Capability inventory: The agent can execute shell commands through the terminal tool to run the fetch and analysis scripts.
Sanitization: Content is normalized for data matching but is not sanitized for potential embedded LLM instructions, which is characteristic of OSINT and research tools that process external data.

osint-investigation