alterlab-ssci-inference-gate
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text (claim sentences) to audit them against statistical rules. While this constitutes a data ingestion surface, the provided Python linter (
scripts/claim_audit.py) uses safe string matching and regular expressions without executing the input or passing it to dangerous functions. There is no evidence of exploitability. - [SAFE]: The included Python script uses only the standard library and does not perform any network operations, file system modifications, or privileged command execution. All operations are local to the agent's environment and align with the stated purpose of auditing academic claims.
Audit Metadata