skills/inference-sh/skills/infsh-cli/Gen Agent Trust Hub

infsh-cli

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructs the user to install the CLI using a piped-to-shell command: curl -fsSL https://cli.inference.sh | sh. This executes a script fetched from the vendor's domain directly in the environment.
  • [EXTERNAL_DOWNLOADS]: The installation process (both automated and manual) involves downloading binary files, manifests, and checksums from dist.inference.sh and cli.inference.sh.
  • [DATA_EXFILTRATION]: The skill documents a 'Local File Uploads' feature where the belt CLI automatically detects local file paths provided in model inputs and uploads them to the vendor's cloud infrastructure. This functionality creates a surface for potential data exposure if an agent is manipulated into passing sensitive local paths (such as keys or configuration files) as input to the tool.
  • Ingestion points: Command-line arguments for the belt app run command (specifically the --input flag described in SKILL.md and references/running-apps.md).
  • Boundary markers: None identified in the provided documentation.
  • Capability inventory: File read and network upload capabilities via the belt CLI tool executed through the Bash tool (restricted to the belt prefix in frontmatter).
  • Sanitization: No specific sanitization or validation of the file paths provided by the agent is described in the documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 07:48 AM
Security Audit — agent-trust-hub — infsh-cli