The Agent Skills Directory

[COMMAND_EXECUTION]: The blockrun_modal tool provides an exec action that allows the agent to execute arbitrary Python code within a hosted sandbox environment. This capability is intended for developer tasks but grants the agent significant autonomy to run scripts.
[EXTERNAL_DOWNLOADS]: The installation instructions direct users to run npx -y @blockrun/mcp@latest, which downloads and executes code from the NPM registry. This is a standard installation method for this vendor's tools.
[PROMPT_INJECTION]: The skill exhibits a significant surface for indirect prompt injection due to its architecture:
Ingestion points: Tools such as blockrun_exa (web search and research papers), blockrun_search (web and news content), blockrun_surf (social media and on-chain data), and blockrun_markets (prediction market events) ingest data from untrusted external sources.
Boundary markers: The instructions do not define boundary markers or provide explicit warnings for the agent to treat data retrieved from these tools as untrusted.
Capability inventory: The agent has the ability to execute arbitrary code via blockrun_modal, make outbound voice calls via blockrun_phone, and perform financial operations via blockrun_wallet.
Sanitization: There are no documented mechanisms for sanitizing or validating external content before it is processed by the agent or passed to other tools.
[REMOTE_CODE_EXECUTION]: The integration of tools that read untrusted web content with a tool that executes arbitrary code (blockrun_modal) creates a potential path for unauthorized remote code execution if an attacker successfully injects instructions into a search result or market listing.

blockrun-mcp-live-data