The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the infsh CLI to perform research and manage user sessions. This access is restricted via the allowed-tools configuration to ensure only the specified CLI can be executed.
[REMOTE_CODE_EXECUTION]: The skill includes an example for generating visualizations by running Python code through a remote executor application (infsh/python-executor). This is a legitimate functional component for creating charts.
[EXTERNAL_DOWNLOADS]: References installation instructions for its required CLI tool on GitHub, which is a well-known and trusted source for documentation and software.
[PROMPT_INJECTION]: The skill ingests data from external search providers (Tavily and Exa), creating a surface for indirect prompt injection. This is standard for research tasks and carries negligible risk in this context.
Ingestion points: External search results from industry benchmarking and research tools (SKILL.md).
Boundary markers: None present.
Capability inventory: Shell access is limited to the infsh CLI toolset (SKILL.md).
Sanitization: Data is processed for inclusion in case study content.

case-study-writing