Reverse Skill Adapter is a thin Codex plugin skill. The reverse-skill repository stays the canonical implementation; the adapter reads its rules, runs its router to pick a primary skill, and requires a validated scope file before any target action.
It does not bundle a second copy of the router. If the repository is not the current workspace and no local path is given, it asks the user to open or clone the repository first. Authorization comes from the repository’s scope contract, not from the plugin or a target name.
What You Get
- A Codex entry point that defers to the repository’s routing configuration
- Router commands for Windows (PowerShell) and for Linux, macOS and Kali (bash)
- A scope gate: a case scope file is created and validated before any target action
- A rule that tools are resolved from the generated tool index and not installed without a request
Common Use Cases
- Starting an authorized reverse engineering or CTF task from Codex inside the reverse-skill repository
- Letting the repository’s routing table choose the primary skill for a task description
- Enforcing a scope file before a security test begins
- Pointing a Codex user to clone the repository when the workspace lacks it
What Makes It Unique
It adds no analysis method of its own. It is an adapter that keeps the repository client-neutral while giving Codex a way in, and it ties authorization to the scope file the repository defines. Merely having the plugin installed or naming a target does not count as permission.
Under The Hood
The skill is a short markdown file with no scripts of its own. It tells the agent to run the repository’s master-route and case-init and case-guard scripts (PowerShell or bash) with the user’s task text, which means the agent runs shell commands, reads repository files and creates case folders and scope files in the workspace. It also reads the generated tool index. It states that it registers no MCP servers and installs no tools unless asked, and it fetches nothing from the web by itself.