deeplearning-ai-agentic-research-agent
Fail
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill directs users to fetch project code from a GitHub repository with the URL
https://github.com/https-deeplearning-ai/agentic-ai-public.git. The organization name 'https-deeplearning-ai' is highly suspicious and appears to be a deliberate attempt to impersonate the official 'deeplearning-ai' organization, posing a major risk of downloading malicious code. - [REMOTE_CODE_EXECUTION]: Instructions provide for building and running a Docker container from the downloaded repository (
docker build -t fastapi-postgres-service .anddocker run). Executing a Docker build and run sequence on code from an impersonating, untrusted source allows for arbitrary code execution with high privileges on the host system. - [COMMAND_EXECUTION]: The skill provides multiple shell commands for the user to execute, including
git clone,docker build, anddocker run. These commands are used to facilitate the installation and execution of the unverified software from the suspicious repository. - [CREDENTIALS_UNSAFE]: Hardcoded database credentials are included in the documentation and connection examples:
postgresql://app:local@localhost:5432/appdb. While these appear to be intended for local development, hardcoding credentials and recommending their use in environment variables in documentation is a poor security practice.
Recommendations
- AI detected serious security threats
Audit Metadata