orca-onboard
Pass
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Downloads project templates and security scripts from the author's official GitHub repository (
github.com/rheinmir/setup). - [COMMAND_EXECUTION]: Utilizes bash and Python to automate codebase scanning, metadata extraction, and file management. It also starts a local background process using
npx serveto host the generated documentation. - [REMOTE_CODE_EXECUTION]: Executes a downloaded installation script (
install-harness.sh) from the vendor's repository. This script is designed to configure security hooks within the agent's environment. - [DATA_EXFILTRATION]: Performs a
git pushoperation to the author's repository at the end of the process to synchronize skill updates. This involves transmitting local configuration data to the vendor's GitHub infrastructure. - [PROMPT_INJECTION]: Analyzes untrusted external project files, creating an attack surface for indirect prompt injection.
- Ingestion points: Source code, configuration files, and READMEs from the analyzed project directory (File:
SKILL.md). - Boundary markers: Absent; the skill does not wrap analyzed content in protective delimiters.
- Capability inventory: Significant capabilities including shell execution, Python execution, and Git network operations (File:
SKILL.md). - Sanitization: No validation or sanitization of input data is performed before analysis.
Audit Metadata