appfactory-builder
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATION
Full Analysis
- EXTERNAL_DOWNLOADS (MEDIUM): The skill requires cloning an external, untrusted repository (https://github.com/MeltedMindz/AppFactory.git). This source is not verified or part of the trusted organization list.
- COMMAND_EXECUTION (MEDIUM): The skill instructs the agent to execute shell scripts and build commands from the downloaded repository, specifically
bash deploy.shandnpm run build. These scripts could perform malicious actions on the local environment. - REMOTE_CODE_EXECUTION (MEDIUM): Uses
npx create-next-app@latestandnpx vercel, which fetch and execute code from the npm registry. While these are common developer tools, automated execution by an agent on untrusted project scaffolds poses a risk of supply chain attack. - DATA_EXFILTRATION (LOW): Commands like
npx vercel --prod --yestrigger the upload of local project files and potentially environment variables to a third-party cloud provider (Vercel) without manual confirmation. - PROMPT_INJECTION (LOW): The
HEARTBEAT.mdfile contains instructions aimed at steering the agent's long-term behavior by establishing a 'heartbeat rotation' of weekly tasks and social media interactions.
Audit Metadata