The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection from the data it processes.
Ingestion points: Untrusted primary sources, such as court filings and news archives, are retrieved from the web using WebFetch and WebSearch.
Boundary markers: The instructions do not define boundary markers (e.g., XML tags or delimiters) to separate fetched content from the agent's instructions, nor do they include warnings to ignore embedded commands.
Capability inventory: The skill has permissions to edit and write files (Write, Edit) and access the network (WebFetch).
Sanitization: The skill lacks mechanisms to sanitize or validate the content of external documents before analysis.
[EXTERNAL_DOWNLOADS]: The skill fetches documents from external repositories, including DocumentCloud, CourtListener, Scribd, and various government agency websites, for investigative purposes.
[COMMAND_EXECUTION]: The skill performs local file system operations to store research findings and configure itself.
It uses Write and Edit tools to create RESEARCH.md and SOURCES.md files within resolved album directories.
It utilizes a load_override function to dynamically incorporate configuration settings from a local research-preferences.md file.

researcher