cors-chain-automation

Fail

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATIONSAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill defines and executes several shell scripts to automate CORS vulnerability discovery. It utilizes curl with custom headers to probe external target domains.
  • [PROMPT_INJECTION]: The cors-poc-generator.sh script is vulnerable to indirect prompt injection (Category 8) where user-provided input is interpolated into generated code.
  • Ingestion points: Target, endpoint, and session variables provided by the user in SKILL.md.
  • Boundary markers: None; inputs are directly embedded into HTML and JavaScript templates.
  • Capability inventory: File-writing via shell redirection and network access via fetch in the generated PoC.
  • Sanitization: None; the skill does not escape or validate input before embedding it into the PoC HTML file.
  • [DATA_EXFILTRATION]: The generated PoC template includes commented-out code for out-of-band (OOB) data exfiltration using a placeholder domain (oastify.com). This demonstrates how a successful CORS exploit can be used to exfiltrate data to an attacker-controlled server.
  • [EXTERNAL_DOWNLOADS]: The verification section performs network requests to httpbin.org, which is a well-known service for testing HTTP clients. These requests are used to confirm the functionality of the probing logic.
Recommendations
  • HIGH: Downloads and executes remote code from: https://httpbin.org/get - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Jul 31, 2026, 04:47 PM
Security Audit — agent-trust-hub — cors-chain-automation