backlink
Fail
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to clone an external repository from a third-party user (
https://github.com/s87343472/backlink-pilot.git) and perform annpm install. This involves downloading and executing arbitrary code from an unverified source on the user's remote Mac Mini. - [COMMAND_EXECUTION]: The skill makes extensive use of
sshto execute shell commands remotely on a Mac Mini andscpto transfer configuration files containing product metadata. This assumes the agent has significant administrative access to the remote machine. - [EXTERNAL_DOWNLOADS]: The skill installs global and local Node.js packages (
bb-browser,js-yaml) from the npm registry and clones source code from GitHub during its setup and execution phases. - [DATA_EXFILTRATION]: The
batch-direct.mjsscript is designed to submit product information—including URLs, descriptions, and contact emails—to a hardcoded list of over 100 external, non-whitelisted directory websites. While this is the intended functionality of the skill, it facilitates large-scale data transmission to third parties. - [REMOTE_CODE_EXECUTION]: The
batch-direct.mjsscript utilizes thebb-browser evalcommand to execute dynamically constructed JavaScript strings within a browser context to automate form interactions, which is a high-risk execution pattern.
Recommendations
- AI detected serious security threats
Audit Metadata