hive-token-diligence
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses specific agent tools (
search_tools,invoke_api_endpoint) to interact with blockchain data providers. No arbitrary shell command execution or unauthorized system access was detected. - [EXTERNAL_DOWNLOADS]: The skill requests network access to fetch live blockchain data (metadata, prices, holder lists) from external providers. This behavior is consistent with its primary purpose of token diligence and uses a restricted toolset.
- [PROMPT_INJECTION]: No patterns of instruction override, jailbreak attempts, or safety filter bypasses were found. The skill emphasizes data provenance and explicitly instructs the agent to avoid answering from model memory for live data.
- [DATA_EXFILTRATION]: No evidence of sensitive local file access or exfiltration of credentials. The data retrieved (public blockchain information) is presented to the user through a structured report template.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external blockchain APIs (token metadata, holder info). While this presents a surface for indirect injection, the risk is minimized by the skill's restricted capabilities (read-only API calls) and the use of a predefined report template for output.
Audit Metadata