The Agent Skills Directory

[SAFE]: The skill operates entirely within the local environment to perform static analysis of binary files. It leverages Python's standard library for all processing, including regex-based pattern matching and PE header parsing. No malicious behaviors or risky dependencies were found. All activities are consistent with the stated purpose of analyzing malware artifacts in a controlled environment.- [PROMPT_INJECTION]: The skill processes untrusted binary data, which serves as a potential surface for indirect prompt injection if the output is consumed by an LLM.
Ingestion points: Binary files are read by scripts/agent.py and the script block in SKILL.md using the read() and read_bytes() methods.
Boundary markers: The skill does not implement specific delimiters to wrap the extracted strings in its final JSON output report.
Capability inventory: Includes file system read access for binary samples and file system write access for saving analysis results to the rust_re directory.
Sanitization: String extraction is constrained to printable ASCII characters and specific length ranges (8-500 characters), which mitigates the risk of processing large or complex malicious payloads.

reverse-engineering-rust-malware