chemical-storage-sorter
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The file
chemical-storage-sorter_audit_result_v2.jsoncontains self-referential assertions regarding the skill's safety and readiness, including fields such as"security": "PASS"and"grade": "Production Ready". This pattern is used to influence automated security analyzers or agent behavior by asserting safety within the content processed by the agent. - Ingestion points:
chemical-storage-sorter_audit_result_v2.jsonis included in the skill package files. - Boundary markers: Absent; the safety claims are presented as valid metadata without delimiters or warnings.
- Capability inventory:
scripts/main.pyperforms local chemical classification based on string matching; no network exfiltration, high-privilege file access, or remote code execution patterns were identified in the source code. - Sanitization: No sanitization, validation, or filtering of these self-referential claims is implemented.
- [METADATA_POISONING]: The presence of
chemical-storage-sorter_audit_result_v2.jsonprovides a deceptive security assessment within the skill metadata, claiming a "Production Ready" status and a "PASS" security verdict, which may mislead users or automated agents regarding the actual risks associated with the skill.
Audit Metadata