The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/agent.py uses subprocess.run to execute an external Python script for JARM scanning. The path to this script is configurable via an environment variable, which could be redirected to an unintended executable.
Evidence: scripts/agent.py lines 78-80: jarm_script = os.getenv("JARM_SCRIPT", "jarm.py") used in subprocess.run(["python3", jarm_script, ...]).
[EXTERNAL_DOWNLOADS]: The skill requires external dependencies including the pyMalleableC2 library and the jarm.py scanner script.
Evidence: SKILL.md and scripts/agent.py references to these external tools.
[PROMPT_INJECTION]: The skill processes potentially untrusted malleable C2 profiles provided by the user, which could contain malicious content designed to influence the agent's analysis or output.
Ingestion points: extract_profile_indicators in scripts/agent.py reads user-supplied file paths.
Boundary markers: None identified in the script logic.
Capability inventory: Network scanning via subprocess execution and report generation via file writing.
Sanitization: No input validation or sanitization is performed on the profile content before parsing.

analyzing-cobalt-strike-malleable-profiles