eastsword-dfyx-code-security-review
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions direct the user to clone a repository from github.com/EastSword, which is not a recognized trusted source.
- [REMOTE_CODE_EXECUTION]: The skill requires downloading external code and installing unverified dependencies via pip, then executing Python scripts such as scripts/pattern_scanner.py and scripts/data_flow_analyzer.py from the cloned repository.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted source code, creating a surface for potential injection of malicious instructions via code comments or strings. 1. Ingestion points: Processes source code across multiple languages provided in the agent context. 2. Boundary markers: The instructions lack any delimiters or warnings to ignore instructions embedded within the code being analyzed. 3. Capability inventory: The skill references scripts that perform file system scanning and data flow analysis. 4. Sanitization: No input validation or sanitization of the code content is documented before processing.
- [METADATA_POISONING]: The skill documentation claims the author is 'ara.so', which contradicts the system-provided author metadata identifying the author as 'reason-machines'.
Recommendations
- AI detected serious security threats
Audit Metadata