x-mentor-skill-nuwa
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data during its account diagnostic workflow in 'Scenario E'.
- Ingestion points: The skill ingests user-provided X Analytics data, CSV data exports, and pasted tweet content as described in
SKILL.md. - Boundary markers: Absent. There are no instructions for the agent to treat ingested data as untrusted or to ignore instructions that might be embedded within user tweets or analytics reports.
- Capability inventory: The skill utilizes file writing capabilities to generate diagnostic reports in .html, .json, and .md formats within the
user-data/directory. - Sanitization: The skill does not define any sanitization, validation, or escaping logic for the external content before it is processed or interpolated into reports.
- [METADATA_POISONING]: There is a discrepancy between the stated skill author and the installation source.
- The skill metadata attributes the work to 'ara.so', while the installation instructions point to an 'alchaincyf' repository, and the context identifies the author as 'reason-machines'. This inconsistency in metadata can be misleading regarding the provenance and maintenance of the skill.
Audit Metadata