The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): The skill's 'Quick Start' instructions include the command curl -fsSL https://cli.inference.sh | sh. This is a critical security risk as it executes a remotely hosted script directly in the user's shell without any prior verification or sandboxing. The domain inference.sh is not among the verified trusted sources.
[EXTERNAL_DOWNLOADS] (HIGH): The installation process fetches binary files from dist.inference.sh. While the documentation claims to perform checksum verification, the process is entirely dependent on the integrity of an untrusted third-party server and script.
[COMMAND_EXECUTION] (MEDIUM): The skill requests broad Bash tool permissions for the infsh command. This grants the agent the capability to perform various actions on the system, including logging into remote accounts and running arbitrary AI applications distributed by the provider.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill creates a surface for indirect prompt injection as it processes and displays output from external LLMs without visible boundary markers or sanitization.
Ingestion points: infsh app run model outputs (SKILL.md).
Boundary markers: Absent.
Capability inventory: Bash(infsh *) (SKILL.md).
Sanitization: Absent.

llm-models