pokemon-player

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the pokemon-agent package directly from an external GitHub repository (github.com/NousResearch/pokemon-agent).- [COMMAND_EXECUTION]: The skill executes shell commands to set up the environment, including pip install and ssh for network tunneling.- [DATA_EXFILTRATION]: The skill establishes an SSH reverse tunnel (ssh -R 80:localhost:9876 nokey@localhost.run) which exposes a local port (9876) to the public internet. This bypasses local firewall protections to provide external access to the emulator dashboard.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted visual data and RAM states from the game emulator.
  • Ingestion points: Visual screenshots are captured via GET /screenshot and processed by a vision model.
  • Boundary markers: No specific delimiters or safety instructions are used to separate game content from agent instructions during vision analysis.
  • Capability inventory: The skill has capabilities for shell execution, network tunneling, and local file access (/home/teknium/pokemon-agent).
  • Sanitization: No sanitization or content filtering is performed on the data ingested from the emulator server.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 05:58 AM