acidrain-security-testing-toolkit
Warn
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill clones a repository from an unknown GitHub account: https://github.com/henry-lewiskpp1107/acidrain-security-script-hub.git.
- [REMOTE_CODE_EXECUTION]: Uses new Function(payload) to test JavaScript execution, which allows for dynamic code evaluation of input data.
- [REMOTE_CODE_EXECUTION]: Employs innerHTML injection within dynamically created iframes to verify script execution.
- [COMMAND_EXECUTION]: Executes several shell commands including git clone, php -S, curl, mkdir, and chmod 755 for environment setup and testing.
- [PROMPT_INJECTION]: The skill is designed to ingest and process arbitrary strings (XSS payloads) and execute them in a browser context. While this is the intended purpose, it lacks sanitization at the ingestion points to facilitate testing, creating a vulnerability surface if used with untrusted data.
- Ingestion points: testPayloads arrays, input parameters in testInputSanitization and testXSSPayload.
- Boundary markers: None present to delimit payloads from instructions.
- Capability inventory: innerHTML injection, new Function execution, iframe creation.
- Sanitization: Explicitly bypassed in several examples to demonstrate reflection.
Audit Metadata