diffing-binary-patches

Warn

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download remote content via a third-party conversion service at https://defuddle.md/. This external dependency is used to fetch and strip HTML from public advisories and reports before processing.
  • [DATA_EXFILTRATION]: By routing external requests through defuddle.md, the skill exposes requested URLs and their contents to a third-party service provider. The documentation explicitly acknowledges that requests leave the local environment and warns against using the service for sensitive or adversary-controlled infrastructure.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests untrusted external data for analysis.
  • Ingestion points: The curl command in SKILL.md fetches content from arbitrary external URLs via defuddle.md.
  • Boundary markers: None. The skill does not define delimiters or instructions to ignore embedded commands within the fetched content.
  • Capability inventory: The skill facilitates binary analysis and vulnerability research using tools like BinDiff and Ghidra.
  • Sanitization: None. Content fetched from remote sources is processed directly by the agent without filtering or escaping.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 7, 2026, 11:27 AM
Security Audit — agent-trust-hub — diffing-binary-patches