The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves data from search engines via tools like tavily/search-assistant and exa/search, which creates a vulnerability to instructions hidden in external web content.
Ingestion points: SKILL.md specifies calls to search assistant tools to gather market data and pain points.
Boundary markers: No specific delimiters or "ignore" instructions are provided to the agent for handling the retrieved search data.
Capability inventory: The agent has access to the Bash tool to execute belt CLI commands and file system operations.
Sanitization: No data validation or sanitization mechanisms are defined for the incoming search results.
[EXTERNAL_DOWNLOADS]: The skill directs the user to fetch installation instructions from an external GitHub repository.
Evidence: The file references https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md for setup.
[COMMAND_EXECUTION]: The skill uses the Bash tool to run commands from a custom CLI utility (belt) to manage research tasks and generate images.
Evidence: Executions include belt login and belt app run commands.

customer-persona