The Agent Skills Directory

[Remote Code Execution] (CRITICAL): The skill explicitly instructs execution of curl -fsSL https://cli.inference.sh | sh. This is a classic piped remote script execution pattern.
Evidence: Found in SKILL.md under the 'Quick Start' section.
Risk: The server at cli.inference.sh can deliver any malicious payload at the moment of execution, leading to complete host takeover.
[External Downloads] (HIGH): The skill relies on external binaries and scripts from an untrusted third-party domain (inference.sh).
Evidence: The skill mentions downloading binaries from dist.inference.sh and verifying checksums.
Risk: While verification is mentioned, the source itself is not a trusted organization (e.g., GitHub, Google, Microsoft), making the software supply chain unverifiable.
[Command Execution] (MEDIUM): The skill requires access to a custom binary (infsh) and uses npx to add more dependencies at runtime.
Evidence: allowed-tools: Bash(infsh *) and npx skills add ... commands.
Risk: Running specialized local binaries or adding unversioned remote skills via npx increases the attack surface and allows for dynamic code loading from unverified sources.

linkedin-content