my-brain-is-full-crew
Fail
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill provides instructions to clone a repository from an untrusted source (https://github.com/gnekt/My-Brain-Is-Full-Crew.git) and execute a bash script (scripts/launchme.sh), which can result in the execution of arbitrary, unverified code on the user's system.\n- [EXTERNAL_DOWNLOADS]: The installation process involves downloading software and configuration files from an external source not identified as a trusted vendor.\n- [COMMAND_EXECUTION]: The skill relies on shell commands for setup and initialization, allowing for the potential execution of malicious commands during the installation phase.\n- [DYNAMIC_EXECUTION]: The installer dynamically populates the .claude/ directory with agent and skill definitions, which are subsequently loaded and executed by the AI agent environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill features agents designed to ingest and process data from untrusted external sources, including emails, calendar invites, and meeting transcripts.\n
- Ingestion points: Data is retrieved from the Gmail and Google Calendar APIs and processed from user-provided meeting notes and recordings.\n
- Boundary markers: The skill documentation does not describe any specific boundary markers or isolation techniques used to distinguish external data from agent instructions.\n
- Capability inventory: The agents are equipped with capabilities to read/write filesystem data in the vault and interact with external network-based APIs.\n
- Sanitization: There is no evidence of input validation or sanitization logic being applied to the external content before it is processed by the AI agents.
Recommendations
- AI detected serious security threats
Audit Metadata