Skills
skills/tul-sh/skills/pitch-deck-visuals/Gen Agent Trust Hub

pitch-deck-visuals

Fail

Audited by Gen Agent Trust Hub on Mar 4, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructions promote the use of curl -fsSL https://cli.inference.sh | sh for installation. This 'pipe to shell' pattern allows an external server to execute arbitrary code on the host system without verification, which is a major security risk for AI agents executing automated tasks.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to run infsh CLI commands, including infsh login and infsh app run. This enables the execution of system-level operations and interaction with remote infrastructure.
  • [EXTERNAL_DOWNLOADS]: The skill fetches and executes applications from inference.sh and fal.ai (e.g., infsh/html-to-image, infsh/python-executor, falai/flux-dev-lora). These are external, non-whitelisted sources that execute logic outside the agent's direct control.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its architecture: 1. Ingestion points: Data provided for pitch deck slides is interpolated into HTML and Python templates used by the infsh tool. 2. Boundary markers: There are no boundary markers or instructions to ignore embedded commands in the interpolated data. 3. Capability inventory: The skill uses the Bash(infsh *) tool to execute remote apps and code. 4. Sanitization: There is no evidence of input validation or escaping before data is processed by the execution tools.
Recommendations
  • HIGH: Downloads and executes remote code from: https://cli.inference.sh - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 4, 2026, 12:29 PM