headless-ghidra-analyze-function
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill performs local binary analysis using ghidra-agent-cli, with all file operations restricted to the local artifacts directory.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted binary metadata which represents a potential injection surface. 1. Ingestion points: Binary symbols and strings (SKILL.md). 2. Boundary markers: Single-quoting of CLI arguments and use of address-based identifiers. 3. Capability inventory: Local file writes and Ghidra CLI execution. 4. Sanitization: Address-based identifier normalization and baseline verification.
Audit Metadata