pokemon-player

Pass

Audited by Gen Agent Trust Hub on Apr 27, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external data from the game's memory state and vision analysis results.
  • Ingestion points: RAM state data (GET /state) and vision analysis of screenshots (vision_analyze).
  • Boundary markers: Absent.
  • Capability inventory: Shell commands for repository cloning, package installation, server management, and SSH tunneling.
  • Sanitization: No explicit sanitization or validation of the ingested game data is mentioned.
  • [EXTERNAL_DOWNLOADS]: Fetches the pokemon-agent package from the vendor's official GitHub repository (github.com/NousResearch/pokemon-agent).
  • [COMMAND_EXECUTION]: Executes shell commands to create virtual environments, install packages in editable mode, and run background processes for the game server and SSH tunnel.
  • [DATA_EXFILTRATION]: Establishes an SSH reverse tunnel via the well-known service localhost.run to expose a local port for an external web dashboard.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 27, 2026, 07:07 AM