trtc
Fail
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: HIGHCREDENTIALS_UNSAFEOBFUSCATIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The file
tools/docsbot.pycontains a hardcoded API key (DOCSBOT_API_KEY) for thedocsbot.aiservice. Hardcoding secrets in skill files is a significant risk as they are visible to users and in log files.\n- [OBFUSCATION]: The fileroom-builder/assets/local-usersig/lib-generate-test-usersig-es.min.jsconsists of minified JavaScript code. While it appears to be a library for generating user signatures, minified code hinders verification of its actual logic during static analysis.\n- [EXTERNAL_DOWNLOADS]: Theruntime/package.jsonfile includes a dependency onpuppeteer. Running this code triggers the download of external browser binaries (Chromium) to the user's environment.\n- [COMMAND_EXECUTION]: Multiple scripts utilize thesubprocessmodule to execute system commands:runtime/telemetry_collector.pyandruntime/lib/platforms.pyrun development tools likeadbandxcrunto capture logs, whiletools/reporting.pyexecutesnodeto process telemetry.\n- [INDIRECT_PROMPT_INJECTION]: The search and documentation retrieval tools (tools/search.py,tools/docs.py,tools/docsbot.py) ingest user-supplied queries and interpolate them into external search requests or KB scans. While telemetry redaction is present, the surface area for processing untrusted data could be used to influence agent behavior through retrieved snippets.
Recommendations
- AI detected serious security threats
Audit Metadata