update-aegis
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill guides the agent to execute local maintenance scripts (
aegis-update.pyandaegis-doctor.py) within the skill's own file structure to perform system checks, registration of new hosts, and software updates. - [EXTERNAL_DOWNLOADS]: The skill provides instructions for the Kimi Code CLI to install a plugin from the author's GitHub repository (
https://github.com/GanyuanRan/Aegis). This is a direct reference to the vendor's own infrastructure and represents the intended update mechanism for that platform. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data by reading local configuration files (
~/.config/aegis/config.toml) and interpreting the JSON output of its own scripts. While these are ingestion points for external data, the use of structured JSON for tool communication is a standard practice that mitigates the risk of instructions being accidentally parsed from data.
Audit Metadata