Reverse Skill

Reverse Skill is a skill router pack for security work driven by an AI coding client. When an agent is handed an APK, a binary, encrypted frontend JavaScript, a capture-the-flag (CTF) challenge, or a pentesting target, the pack routes the task to a matching methodology, checks which tools are available, and runs a repeatable workflow rather than guessing commands.

The problems it targets are stated plainly: an agent does not know whether a task calls for jadx, apktool, Frida, IDA, or Burp Suite, different artifact types each need their own playbook, tools and MCP servers are scattered across machines, and past experience is not reused. Routing is therefore driven by one structured configuration file, with a fast ladder for the common cases and a full matrix behind it.

A scope gate runs before any action against a target. A case directory records authorization and a network profile, then work proceeds through an evidence, finding, and path chain into a timeline, a report, and a field journal. Case review is read-only and includes artifact fixity checks.

The routing core, regression suite, manifests, and case workflow do not depend on a specific AI client. Claude Code, Codex, Cursor, and OpenCode load the repository through their own adapter or project-instruction mechanism, and client-specific configuration stays outside the routing contract. An optional adapter plugin for Codex delegates to the same routing core and registers no external MCP servers automatically. The project states it is intended only for lawful security research, education, CTF competitions, and testing of systems you own or are explicitly authorized to assess.