skills/jablonkai/skills/krita/Gen Agent Trust Hub

krita

Warn

Audited by Gen Agent Trust Hub on Oct 11, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONDYNAMIC_EXECUTIONPERSISTENCECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The bridge server (scripts/krita_bridge/krita_bridge.py) listens on 127.0.0.1:8737 and executes arbitrary Python scripts received via HTTP POST. This allows any local user or process to execute code in the context of the user running Krita without any authentication (e.g., token or password).
  • [DYNAMIC_EXECUTION]: The _run function in scripts/krita_bridge/krita_bridge.py uses the dangerous exec() and compile() functions to run code received from the network.
  • [PERSISTENCE]: The scripts/krita-install-bridge.sh script installs the bridge as a Krita Python plugin (pykrita) and modifies the application's configuration file (kritarc). This ensures the code-execution bridge starts automatically whenever Krita is launched.
  • [COMMAND_EXECUTION]: The skill includes shell scripts (scripts/krita-install-bridge.sh, scripts/krita-send.sh) that use command-line tools like curl, pgrep, and cp to manage the bridge and modify system-level application settings.
  • [DATA_EXFILTRATION]: Because the bridge captures and returns all stdout and stderr from the scripts it runs, a malicious script could easily read sensitive local files, credentials, or session data and transmit them back to the caller.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 11, 2026, 03:50 PM