blockrun
Warn
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of a third-party Python package 'blockrun-llm' via pip, which is used to manage the wallet and interface with external LLM providers.
- [COMMAND_EXECUTION]: The skill uses Python commands to initialize an agent wallet, generate QR codes, and execute chat requests with external models.
- [DATA_EXFILTRATION]: User prompts and context are transmitted to the BlockRun service and subsequently to external providers (OpenAI, xAI, DeepSeek, Google) to fulfill capabilities not natively supported by the agent.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from live X/Twitter searches and responses from external LLMs without specific sanitization instructions.
Audit Metadata