The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation includes a command to install the Inference CLI using a pipe-to-shell pattern: curl -fsSL https://cli.inference.sh | sh. This executes unverified remote code directly in the user's environment.
[EXTERNAL_DOWNLOADS]: The skill uses the npx tool to dynamically download and add remote skills from the inference-sh/skills repository at runtime.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute infsh commands, which interact with external cloud services to run applications and manage configurations.
[REMOTE_CODE_EXECUTION]: The skill facilitates the execution of remote AI models (e.g., falai/flux-dev-lora, infsh/agent-browser) by passing user-defined arguments to the infsh app run command.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes untrusted user data for changelogs and visual generation and passes it directly to tool execution.
Ingestion points: User-provided changelog text and visual prompts in SKILL.md examples.
Boundary markers: Absent; there are no delimiters or instructions to ignore embedded commands in the processed data.
Capability inventory: The skill has access to the Bash tool with infsh capabilities, enabling network operations and remote app execution.
Sanitization: Absent; the skill does not implement validation or escaping for the strings interpolated into the infsh command arguments.

product-changelog