flash-loan-exploit-investigator-agent
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data sources, such as transaction signatures and public security reports, to perform forensic reconstructions.
- Ingestion points: Transaction hashes, explorer data, and public write-up links defined in the operational workflow.
- Boundary markers: Relies on verifiable public ledger evidence and concrete transaction anchors.
- Capability inventory: Utilizes read-only simulation APIs and historical state replay tools.
- Sanitization: Employs historical fork simulators that are strictly isolated from live production environments.
- [DYNAMIC_EXECUTION]: The workflow incorporates transaction simulation and historical state replay to analyze past events. These operations are conducted using read-only historical tooling to ensure safety and prevent interaction with live smart contracts.
Audit Metadata