frida-17 — Claude Code skill: Frida 17 script compatibility

What frida-17 does

This skill tracks what broke in Frida 17.0.0 (May 2025). The static Module methods are gone: Module.findBaseAddress, Module.getBaseAddress, Module.findExportByName and Module.enumerateExports no longer run. You go through Process.findModuleByName or Process.getModuleByName and then call methods on the module itself — lib.base, lib.findExportByName('open'), lib.enumerateExports(). Static Memory methods went the same way; reads and writes now hang off the pointer: ptr.readU32(), ptr.writeU32(value).

Naming is the other trap. hexdump, ptr and NULL belong to Frida, so a custom function hexdump(ptr, len) dies with TypeError: cannot define variable 'hexdump'. The skill suggests dumpHex instead. It also catches toUInt32(), which never existed — for sizes under 2^31 use toInt32(). The callback style for module enumeration (onMatch/onComplete) and enumerateModulesSync() are removed, and Process.enumerateModules() hands back an array directly.

A 10-point checklist covers script review, plus working patterns: polling for a library with setInterval, hooking libc through Interceptor.attach, a custom hex dump. The Java Bridge is unchanged, but a Java byte[] cannot go straight into writeByteArray() — you loop over the bytes. Java hook examples use .overload() and walk .overloads.