The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill documents the use of sudo to attach to running processes using strace -p and to execute dtruss on macOS. This is a standard requirement for these tools to function across process boundaries for monitoring.
[EXTERNAL_DOWNLOADS]: The skill suggests downloading tools from external sources, including rohitab.com (for API Monitor) and using the Chocolatey package manager (choco install sysinternals). These are established resources in the reverse engineering community.
[INDIRECT_PROMPT_INJECTION]: The skill involves the agent processing log files generated by external tools, which could contain malicious instructions if the traced target is designed to subvert the analysis.
Ingestion points: The skill instructions suggest reading and analyzing logs like trace.log and libcall.log generated from tool output (SKILL.md, references/commands.md).
Boundary markers: The instructions do not define specific delimiters or guardrails for the agent when interpreting the content of these logs.
Capability inventory: The skill includes instructions to install software packages and execute system-level tracing commands with elevated privileges.
Sanitization: There is no evidence of sanitization or validation of the log data before it is presented to the agent for analysis.

re-tracing