The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (HIGH): The skill contains a manual installation command curl -fsSL https://cli.inference.sh | sh. This pattern is highly dangerous as it executes unverified code from an external server directly in the shell without any integrity checks or version pinning.
[COMMAND_EXECUTION] (MEDIUM): The skill requests permission for Bash(infsh *). Since infsh is the tool installed via the unverified remote script mentioned above, any vulnerability or malicious payload in that script gains command execution privileges on the user's system.
[EXTERNAL_DOWNLOADS] (MEDIUM): The documentation includes instructions to use npx skills add inference-sh/skills@.... These are external dependencies hosted outside of the trusted organizations list (e.g., Anthropic, OpenAI, Microsoft), posing a risk of supply chain attack if the inference-sh namespace is compromised.
[PROMPT_INJECTION] (LOW): The skill is designed to ingest external data (PR descriptions, feature notes) to generate changelogs. This creates a surface for Indirect Prompt Injection (Category 8).
Ingestion points: Data provided via prompt for changelog generation.
Boundary markers: None identified in the provided templates.
Capability inventory: Bash execution via infsh, image generation via remote APIs.
Sanitization: No evidence of input sanitization or escaping before processing.

product-changelog