The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill recommends installing its CLI tool using a piped shell command: curl -fsSL https://cli.inference.sh | sh. This executes a remote script directly in the user's shell environment. This is a vendor-provided installer for the primary tool associated with the skill.
[COMMAND_EXECUTION]: Numerous examples demonstrate the use of shell commands and scripts for automation, including:
Process management using wait, $!, and background execution &.
File system operations such as mkdir, cat, and output redirection > /path/to/file.
System utilities including date, sleep, and echo.
Persistence via crontab configuration for periodic task execution.
[EXTERNAL_DOWNLOADS]: The skill references several external resources:
CLI installer at https://cli.inference.sh and binary downloads from dist.inference.sh.
Extension of capabilities via npx skills add targeting the inference-sh/skills repository.
[PROMPT_INJECTION]: The skill contains patterns that create an attack surface for indirect prompt injection (Category 8):
Ingestion points: Processes external data from command-line arguments ($INPUT_TEXT in the Conditional Workflow pattern) and local file contents ($(cat $file) in the Data Processing Pipeline).
Boundary markers: Absent; the external data is directly interpolated into the JSON-formatted prompt strings.
Capability inventory: The skill utilizes the infsh tool to send data to various LLM providers and curl to send alerts to external webhooks.
Sanitization: No escaping, filtering, or validation is performed on the ingested data before it is sent to the models.

ai-automation-workflows