The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the user to install a third-party CLI tool (belt-sh/cli) and execute remote applications (e.g., falai/flux-dev-lora, tavily/search-assistant) hosted on the inference.sh platform. These applications execute logic on remote infrastructure that is not within the pre-approved trusted list.
[EXTERNAL_DOWNLOADS]: The skill points users to an external installation script and guidelines hosted on GitHub (inference-sh/skills) to set up the necessary environment.
[COMMAND_EXECUTION]: The skill uses the belt command via the Bash tool to perform various tasks. This includes belt login for credential management and belt app run for triggering remote tasks.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests data from external search tools (tavily/search-assistant and exa/search).
Ingestion points: Data enters the context from the outputs of the search assistant and Exa search commands in SKILL.md.
Boundary markers: None identified; the agent is instructed to use the results for preparation and strategy without explicit isolation of the untrusted content.
Capability inventory: The skill has access to the Bash tool and can execute shell commands via belt.
Sanitization: There is no evidence of sanitization or filtering applied to the external search data.

product-hunt-launch