The Agent Skills Directory

[SAFE]: The skill performs legitimate graphics analysis and modification tasks. The included Python scripts (tim2.py, widescreen.py, gspacket.py) rely entirely on the Python standard library (such as struct, zlib, and argparse) to handle binary file formats. No evidence of credential harvesting, network exfiltration, or unauthorized command execution was detected.
[INDIRECT_PROMPT_INJECTION]: While the skill processes untrusted binary data from user-provided game files, the risk is negligible due to the nature of the processing.
Ingestion points: Binary data is ingested from local game files (ELFs, TIM2 textures) and RAM dumps (console/dumps/*.bin).
Boundary markers: Not applicable, as the scripts perform structured binary parsing rather than natural language interpretation of the file content.
Capability inventory: The scripts are limited to local file system operations (reading and writing textures/patches) and do not have network access or the ability to execute arbitrary shell commands.
Sanitization: The tools use strict binary schema validation (magic bytes, header sizes, and fixed-width fields) to process data, preventing malicious content in the files from influencing agent behavior.

console-graphics