sn-ppt-workbench
Fail
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: HIGHDATA_EXFILTRATIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [DATA_EXFILTRATION]: Accesses sensitive configuration files. The script
scripts/open_workbench.pyreads~/.hermes/.envand other platform-specific environment files to extract authentication secrets such asAPI_SERVER_KEYandWORKBENCH_GATEWAY_API_KEY. - [COMMAND_EXECUTION]: Executes code from unverified local paths. The skill implements a search heuristic in
scripts/open_workbench.pythat identifies and runs Node.js scripts found in user directories like~/Repository/usingsubprocess.run(). - [COMMAND_EXECUTION]: Spawns detached background processes. The Node.js launcher
workbench-runtime/bin/sensenova-ppt-workbench.mjsusesspawnwithdetached: trueto run a server in the background, which can be used to maintain persistence of unauthorized processes. - [CREDENTIALS_UNSAFE]: Hardcoded default credential. The launcher code in
sensenova-ppt-workbench.mjscontains a hardcoded fallback API keyppt-editor-devused for session authentication.
Recommendations
- AI detected serious security threats
Audit Metadata