The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous pre-configured shell commands utilizing curl to exploit high-severity vulnerabilities (CVEs) on enterprise VPN hardware. This includes payloads for path traversal (CVE-2018-0296, CVE-2020-3452, CVE-2018-13379, CVE-2019-11510) and pre-authentication Remote Code Execution (CVE-2024-3400, CVE-2024-21887).
[CREDENTIALS_UNSAFE]: The skill contains a reference table of hardcoded default usernames and passwords for major VPN vendors including Cisco (admin/cisco), Fortinet (admin/), Citrix (nsroot/nsroot), Palo Alto (admin/admin), Pulse Secure (admin/password), F5 (root/default), and SonicWall (admin/password).
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external targets, creating a vulnerability surface for indirect prompt injection.
Ingestion points: HTTP responses and metadata retrieved via curl from arbitrary external targets in SKILL.md.
Boundary markers: None identified; instructions do not include delimiters or warnings for the agent to ignore embedded instructions in target output.
Capability inventory: The skill utilizes curl for network requests, python3 for payload generation, and references nuclei for automated vulnerability scanning.
Sanitization: No sanitization, filtering, or validation logic is provided for the data retrieved from external targets before it is processed by the agent.
[DYNAMIC_EXECUTION]: The skill uses inline Python commands (python3 -c) to dynamically generate large memory-disclosure payloads (e.g., for the 'Citrix Bleed' CVE-2023-4966 exploit) which are then interpolated into shell commands.

enterprise-vpn-attack