The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of Bash scripts to orchestrate AI tasks, utilizing the vendor-provided infsh CLI tool. It also provides examples for modifying system cron jobs to establish scheduled automation cycles.
[EXTERNAL_DOWNLOADS]: Contains instructions to download and install supplementary skills from the vendor's repository using the npx skills add command. These resources are hosted within the vendor's own namespace.
[PROMPT_INJECTION]: The 'Data Processing Pipeline' example demonstrates a pattern susceptible to indirect prompt injection. By using $(cat $file) to interpolate raw file content directly into an LLM prompt without delimiters or sanitization, the agent may inadvertently follow instructions contained within the processed data files.
Ingestion points: data_processing.sh reads local data from the ./data/raw directory.
Boundary markers: Absent; data is appended directly to the prompt string.
Capability inventory: Executes external AI models via the infsh CLI.
Sanitization: No escaping or validation is performed on the file content before processing.

ai-automation-workflows