steam-workshop-helper

Pass

Audited by Gen Agent Trust Hub on Jul 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides shell commands using curl.exe to interact with the official Steam Web API (api.steampowered.com). These commands are used to retrieve item details and query workshop content.
  • [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by processing external data (titles and descriptions) from the Steam Workshop. This data is authored by third-party modders and is ingested into the agent's context during queries.
  • Ingestion points: Data enters the context via curl command outputs and the steam-mcp tool responses containing workshop item metadata from SKILL.md.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the retrieved workshop metadata as untrusted or to ignore embedded instructions.
  • Capability inventory: The agent has the capability to execute shell commands (curl.exe) and interact with the provided MCP tools.
  • Sanitization: No sanitization or validation logic is defined for the external content retrieved from the Steam API.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 26, 2026, 08:02 PM
Security Audit — agent-trust-hub — steam-workshop-helper