nansen-token-forensics
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the
nansencommand-line utility to perform token transfers, flows, and flow-intelligence research. This tool is a legitimate resource associated with the skill's author, nansen-ai. - [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection through the ingestion of external blockchain data.
- Ingestion points: Data returned from the Nansen API via the CLI, particularly the
from_address_labelandto_address_labelfields, which are sourced from external blockchain metadata. - Boundary markers: The skill does not implement explicit delimiters or instructions to treat the command output strictly as data, which could allow malicious labels to influence agent behavior.
- Capability inventory: The skill allows the agent to execute shell-based research commands and interpret the results.
- Sanitization: No evidence of sanitization, filtering, or validation of the returned labels or blockchain data is present in the skill code.
Audit Metadata