skills/mukul975/anthropic-cybersecurity-skills/performing-malware-triage-with-yara/Gen Agent Trust Hub
performing-malware-triage-with-yara
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to download community YARA rules from well-known repositories on GitHub (Yara-Rules and Neo23x0/signature-base).
- [COMMAND_EXECUTION]: The workflow involves local execution of the YARA and YARA-C command-line tools for scanning files and compiling rule sets.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to triage untrusted files, which represents an ingestion surface. Ingestion points: Files are read from a user-specified directory by the
scripts/agent.pyscript. Boundary markers: No explicit prompt boundary markers are used as the data is processed by the pattern-matching engine rather than interpolated into an LLM prompt. Capability inventory: The skill is limited to local file system reading and writing JSON reports; no network exfiltration or privileged execution capabilities were detected. Sanitization: Extracted strings are decoded using UTF-8 with error replacement and are truncated to 80 characters.
Audit Metadata