The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides functional shell commands using curl, openssl, and printf to probe and exploit Host Header vulnerabilities. These include sophisticated techniques for Host smuggling, routing-based SSRF to access internal services (e.g., EC2 metadata service), and bypassing edge access control lists (ACLs).
[DATA_EXFILTRATION]: The skill instructs the agent to facilitate the exfiltration of sensitive data, specifically password reset tokens and OAuth authorization codes, to attacker-controlled infrastructure (referenced as evil.com or via a Burp Collaborator domain).
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it is designed to ingest and process data from external web servers. Evidence chain:
Ingestion points: Output from network commands (curl, openssl) interacting with the $TARGET server in Phase 1-5.
Boundary markers: The skill contains a 'Mandatory confirmation gate' requiring human verification of the target and scope, which acts as a manual boundary.
Capability inventory: The skill uses curl and openssl for network communication and grep for string reflection analysis.
Sanitization: No automated sanitization or filtering is applied to the response content retrieved from the target servers before being processed by the agent.

hunt-host-header