The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends installing the belt-sh/cli package via npx and refers to external installation documents on GitHub (inference-sh/skills).
[REMOTE_CODE_EXECUTION]: The skill instructs the user to install an external package (belt-sh/cli) and run remote applications using the belt tool, which involves executing code from third-party sources.
[COMMAND_EXECUTION]: The skill provides instructions to run the belt CLI tool for logging in and executing various remote applications, which involves running shell commands.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from external search applications like Tavily and Exa.
Ingestion points: Data retrieved from web searches by tavily/search-assistant and exa/answer (SKILL.md).
Boundary markers: Absent; there are no specific markers used to delimit untrusted search results from the rest of the prompt context.
Capability inventory: The skill has the capability to run the belt CLI tool and install external skills (SKILL.md).
Sanitization: Absent; the skill does not specify any filtering or sanitization steps for the data received from external tools.

landing-page-design