The Agent Skills Directory

[COMMAND_EXECUTION]: The skill describes using common system and security tools (file, strings, checksec, upx, ltrace, strace, gdb) for local binary analysis. These commands are standard for the described purpose and do not involve unauthorized access or exfiltration.
[DYNAMIC_EXECUTION]: Provides a Python code template for the Z3 solver to automate constraint solving. The snippet is a static example and does not dynamically execute untrusted input or remote code.
[INDIRECT_PROMPT_INJECTION]: The methodology involves processing external, potentially untrusted binary files. Ingestion points: External binaries (ELF/PE/exe) provided as CTF handouts. Boundary markers: None; binaries are treated as raw data for analysis. Capability inventory: Includes tools for code execution and file modification (gdb, upx) used within the local analysis environment. Sanitization: Standard analysis tools are used; no additional sanitization is implemented for the binary content itself.

ctf-rev