The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/agent.py executes the Volatility 3 framework using subprocess.run. It correctly passes command arguments as a list and utilizes shlex.split for supplemental arguments, which effectively prevents shell injection vulnerabilities.
[CREDENTIALS_UNSAFE]: The skill provides functionality to extract password hashes, LSA secrets, and cached credentials from volatile memory. This capability is the intended primary purpose of the forensic tool and is documented for use in incident response scenarios.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and displays data from memory dumps, which are untrusted external sources from potentially compromised systems.
Ingestion points: Results from plugins like windows.pslist and windows.cmdline in scripts/agent.py.
Boundary markers: The skill does not currently use specific delimiters to separate forensic output from agent instructions.
Capability inventory: The skill can execute subprocesses, read files, and extract system artifacts.
Sanitization: Artifacts such as command-line strings are retrieved directly from memory without additional escaping.
[METADATA_POISONING]: A minor discrepancy is noted between the author field in the SKILL.md frontmatter ('mahipal') and the copyright holder in the LICENSE file ('mukul975'), though this appears to be a naming variation rather than a deceptive practice.

analyzing-memory-dumps-with-volatility