parallel-recon-triad

MALICIOUS. This skill is purpose-built to let an AI agent conduct persistent, stealthy, self-improving offensive reconnaissance and exploitation-oriented probing against external targets. The install steps are mostly ordinary distro packages, but the skill’s actual footprint—autonomous attack orchestration, WAF bypass, OPSEC concealment, third-party routing, and credential-use escalation—is fundamentally incompatible with a benign developer-assistant purpose.

This prompt template is high-risk attack-orchestration content. It instructs an autonomous, persistent agent to perform stealthy, large-scale reconnaissance against third-party sites, including enumeration of WordPress/XMLRPC surfaces, probing for exposed secrets, downloading JS bundles to extract API keys/JWTs, and attempting access to highly sensitive files/backup artifacts (e.g., /.env, /.git, wp-config backups, logs/backups). Combined with Tor/proxy circuit rotation and WAF/evasion logic, it is best treated as hostile tooling/payload rather than benign software.