ai-automation-workflows
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill heavily utilizes shell commands and script templates (
bash,python) to automate tasks using thebeltCLI. These are standard for its stated purpose of building AI workflows. - [INDIRECT_PROMPT_INJECTION]: The
data_processing.shscript reads raw content from local.txtfiles and interpolates it directly into an LLM prompt using$(cat $file). This pattern is vulnerable to indirect prompt injection if the source files contain malicious instructions. - Ingestion points: The
data_processing.shscript reads files from./data/rawusingcat. - Boundary markers: No boundary markers or delimiters are used to separate the file content from the system instructions in the prompt.
- Capability inventory: The skill uses
belt app runwhich executes remote AI models andBash(belt *)for local execution. - Sanitization: No sanitization or escaping is performed on the file content before it is embedded in the prompt.
Audit Metadata