gbrain-upgrade
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: Executes the
gbraincommand-line utility to check for updates, read/write configuration settings, and apply software upgrades. - [REMOTE_CODE_EXECUTION]: Utilizes
gbrain self-upgradeto download and install new versions of the software. This is a controlled vendor-provided update mechanism and is the primary purpose of the skill. - [PROMPT_INJECTION]: Implements defensive instructions to ignore third-party commands that might be found in upgrade markers, ensuring the agent only executes the trusted, hardcoded upgrade command.
- [INDIRECT_PROMPT_INJECTION]: Processes external JSON data containing changelogs. The skill mitigates risks associated with untrusted data by instructing the agent to summarize content into a fixed bullet-point format rather than rendering raw external strings.
Audit Metadata