gstack-upgrade
Warn
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill downloads content from an external repository (https://github.com/garrytan/gstack.git) and subsequently executes a script named ./setup from the downloaded files. This pattern allows the remote source to determine the code executed on the local system.\n- [EXTERNAL_DOWNLOADS]: The skill uses git clone and git fetch to retrieve updates from a remote GitHub repository to facilitate the upgrade process.\n- [COMMAND_EXECUTION]: The skill extensively uses shell commands to manage files, detect installation paths, handle git state, and perform the upgrade process, including moving directories and running installation scripts.\n- [PROMPT_INJECTION]: The skill ingests untrusted data from the CHANGELOG.md file in the downloaded repository (Step 6) and incorporates it into the agent's output. This creates a surface for indirect prompt injection where malicious instructions in the changelog could influence the agent's behavior during summarization. Ingestion points: CHANGELOG.md; Boundary markers: None; Capability inventory: Bash, Write, AskUserQuestion; Sanitization: None.
Audit Metadata