gate-mcp-openclawinstaller
Pass
Audited by Gen Agent Trust Hub on Mar 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
install.shscript executes various shell commands to detect dependencies, manage server configurations via themcporterCLI, and verify successful installations. - [EXTERNAL_DOWNLOADS]: The installer uses
npx -y gate-mcpto fetch and execute the Gate trading server directly from the NPM registry. It also configures several HTTP-based MCP servers hosted atapi.gatemcp.ai, which is consistent with the vendor's infrastructure for these tools. - [PROMPT_INJECTION]: The skill provides tools for retrieving market tickers and news feeds (e.g.,
gate-info.list_tickers,gate-news.list_news). This functionality introduces a surface for indirect prompt injection, where external data from news sources or market updates could contain instructions intended to influence the agent's behavior. - Ingestion points: Tools
gate-infoandgate-newsfetch external content from public market feeds and news APIs. - Boundary markers: None identified in the provided instructions; the agent processes the retrieved text directly.
- Capability inventory: The skill includes capabilities for trading and wallet operations (
gate,gate-wallet) which could be targeted if the agent is manipulated via malicious data ingestion. - Sanitization: The underlying server logic is responsible for data sanitization, which is not visible in the installer scripts or instructions.
Audit Metadata