The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the belt-sh/cli package via npx and provides links to external installation scripts hosted on GitHub (e.g., https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md). It also suggests adding further skills from the inference-sh organization.
[REMOTE_CODE_EXECUTION]: The skill uses the belt app run command to execute remote logic, such as infsh/html-to-image and falai/flux-dev-lora. This involves sending data to and executing code on remote infrastructure provided by third-party services.
[COMMAND_EXECUTION]: The skill utilizes shell commands (belt login, belt app run, npx skills add) to interact with the environment and external services. These commands are necessary for the skill's primary function but involve significant capabilities.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted data by interpolating HTML and prompts into remote execution commands.
Ingestion points: HTML strings and image prompts provided to belt app run in SKILL.md.
Boundary markers: None identified; untrusted content is passed directly as command arguments.
Capability inventory: Execution of shell commands and remote applications via the belt CLI.
Sanitization: No evidence of input sanitization or validation before passing content to remote tools.

og-image-design