> For the complete documentation index, see [llms.txt](https://tranchevest.gitbook.io/tranchevest-1/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://tranchevest.gitbook.io/tranchevest-1/risk-assessment-agent-framework.md).

# Risk Assessment Agent Framework

<figure><img src="/files/NF2LGOLZseVDToqtbIgr" alt=""><figcaption></figcaption></figure>

* **External Data Sources** (Blue) - APIs for market data, sentiment analysis, and exchanges
* **Data Processing Layer** (Green) - ETL pipelines, sentiment analysis, time series analysis, and data fusion
* **AI Model Layer** (Orange) - The three main agents:
  * Sheriff: Core risk assessment engine
  * Prism AI: Natural language interface with LORA and Claude
  * Ace Agent: Recommendation engine
* **Middleware Layer** (Pink) - A2A message bus, MCP protocol handler, API gateway, and auth service
* **Application Layer** (Purple) - Various user interfaces including web, Twitter, Telegram, and portfolio dashboard