alpaca-trading-mcp

Fail

Audited by Gen Agent Trust Hub on May 17, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's installation documentation instructs users to download and execute an installation script for the 'uv' package manager from 'astral.sh', which is the official domain for the tool's developers.
  • [REMOTE_CODE_EXECUTION]: The skill configuration utilizes 'uvx' to dynamically download and run the 'alpaca-mcp-server' package from the Python package registry. This is the standard method for deploying this MCP server.
  • [PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection by design.
  • Ingestion points: Untrusted external data is brought into the agent's context through tools like 'get_news', which fetches market news articles, and 'get_corporate_actions_announcements' (SKILL.md).
  • Boundary markers: No specific delimiters or instructions to ignore embedded commands within the fetched market data are provided in the instructions.
  • Capability inventory: The agent is granted highly sensitive capabilities, including the ability to place financial trades via 'post_orders' and liquidate assets via 'delete_positions_by_symbol_or_asset_id' (SKILL.md).
  • Sanitization: The skill does not document any mechanisms for sanitizing or validating the content of the external news or market data before it is processed by the agent.
Recommendations
  • HIGH: Downloads and executes remote code from: https://astral.sh/uv/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
May 17, 2026, 10:50 PM
Security Audit — agent-trust-hub — alpaca-trading-mcp