The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and analyze untrusted external data, which is an inherent part of CTF solving and game reversing. This creates a surface for indirect instructions to be processed by the agent.
Ingestion points: Processes game binaries (.exe, ELF), .NET assemblies (Assembly-CSharp.dll), native libraries (GameAssembly.dll), Unity asset files (*.assets, *.dmp), memory dumps, and network captures (.pcap).
Boundary markers: No explicit delimiters or instructions to ignore embedded content are used when processing external strings or data.
Capability inventory: The skill environment allows for subprocess execution of reversing tools, file system manipulation, direct memory access via /proc/pid/mem, binary patching, and network socket communication.
Sanitization: No sanitization is performed on ingested data before it is processed or used in analysis output.
[EXTERNAL_DOWNLOADS]: The skill downloads and installs several well-known security research tools from official or established sources.
Fetches the ilspycmd utility via the official .NET tool installer.
Clones the AssetRipper project from its public GitHub repository.
References and provides links to established tools such as Il2CppDumper, FModel, and gdsdecomp for specific reversing workflows.
[COMMAND_EXECUTION]: The skill makes extensive use of standard command-line utilities and script-based automation for binary analysis.
Employs strings, grep, file, ls, and checksec to perform initial triage on target artifacts.
Uses Python scripts to automate memory manipulation (reading/writing to /proc/pid/mem), binary patching of comparison logic, and network packet replay.
Utilizes engine-specific tools like dnSpy, Il2CppDumper, and godotpcktool for deep analysis.

game-ctf