modifying-the-harness
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides tools and instructions to configure 'command hooks.' These hooks allow shell commands to be executed automatically in response to platform events like
PreToolUseorSessionStart, establishing event-driven command execution triggers. - [SAFE]: External communication is performed via
curlcommands targeting the vendor's official API (api.letta.com). This is used for legitimate agent configuration tasks such as updating the agent's name, description, or model settings. - [SAFE]: The included scripts (
add_hook.py,add_permission.py, andshow_config.py) are focused on managing local configuration JSON files located in the standard~/.letta/directory and project-local directories. - [SAFE]: Although the skill enables the modification of the agent's own security policy and permission rules, this is the explicitly documented purpose of the skill provided by the vendor.
Audit Metadata