b2-object-lock
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata (such as bucket names and object keys) retrieved from the B2/S3 environment. Malicious instructions embedded in this metadata could theoretically influence the AI's behavior. However, the risk is mitigated by explicit safety gates that require human confirmation for all state-changing or destructive tools. \n
- Evidence: \n
- Ingestion points: Metadata is ingested from
b2_list_buckets,s3_list_object_versions, ands3_head_objectas described inSKILL.md. \n - Boundary markers: The skill lacks explicit boundary markers or 'ignore' instructions for the metadata it processes. \n
- Capability inventory: The skill utilizes powerful tools such as
b2_update_bucket,b2_update_file_retention,b2_update_file_legal_hold,s3_delete_object, ands3_delete_objects. \n - Sanitization: There is no evidence of sanitization or filtering of the retrieved metadata before it is used in the model's context.
Audit Metadata