The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation recommends installing the infsh CLI using curl -fsSL https://cli.inference.sh | sh. This pattern involves piping a remote script directly into a shell, which allows for the execution of arbitrary code on the host machine. While the script is hosted on the vendor's domain, this method remains a potential security risk.- [EXTERNAL_DOWNLOADS]: The skill's installation process downloads binary files and checksums from dist.inference.sh.- [COMMAND_EXECUTION]: The skill requests Bash(infsh *) permissions, which allows the agent to execute any available command within the infsh CLI toolset.- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its core functionality.
Ingestion points: Data enters the agent's context from untrusted external websites via the tavily and exa search and extraction apps mentioned in SKILL.md.
Boundary markers: Examples show data being piped directly into further tools without delimiters or instructions to ignore embedded prompts.
Capability inventory: The agent possesses the Bash(infsh *) capability, which can be used to execute network requests and trigger other apps.
Sanitization: There is no evidence of content sanitization or validation for the data retrieved from external sources.

web-search