The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation includes an installation command curl -fsSL https://cli.inference.sh | sh which downloads and executes a script from a remote server without verification. This is a high-risk pattern that allows a third-party server to execute arbitrary code on the user's system.
[COMMAND_EXECUTION]: The skill requires the Bash tool with a wildcard permission for the infsh command. This grants the agent significant power to execute system-level commands, which could be abused if the agent is compromised or misled by external input.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to how it handles web content extraction.
Ingestion points: The skill uses tavily/extract and exa/extract apps to fetch text and data from external URLs.
Boundary markers: There are no boundary markers or instructions to treat the extracted web content as untrusted data.
Capability inventory: The agent has access to the host system via the Bash tool and the infsh CLI.
Sanitization: The skill lacks any mechanism for sanitizing or filtering the content retrieved from the web before it is passed to the language model for analysis.

web-search