infsh-cli
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill provides installation instructions for the vendor's command-line tool using a pipe-to-shell pattern:
curl -fsSL https://cli.inference.sh | sh. This script downloads the binary from the vendor's distribution infrastructure. - [COMMAND_EXECUTION]: The skill documentation includes numerous examples of executing the
beltCLI tool to perform tasks such as application discovery, deployment, and configuration of shell completions in system or user directories. - [DATA_EXFILTRATION]: The CLI tool includes a feature that automatically detects local file paths provided in application inputs and uploads the referenced files to the vendor's cloud platform for processing. This is documented as a core capability for multi-modal AI workflows.
- [PROMPT_INJECTION]: The skill interacts with external applications and search engines (such as Tavily and Exa) whose outputs are ingested into the agent's context. This creates a surface for indirect prompt injection as there are no explicit sanitization steps or boundary markers defined in the instructions for handling these external outputs.
Audit Metadata