Agent Connectors
How your AI agent talks to Jetro.
What are agent connectors?
Agent connectors are the bridge between your AI agent and Jetro’s capabilities. When you install Jetro, it exposes a set of tools using the MCP (Model Context Protocol) standard. Any MCP-compatible agent can discover and use these tools automatically.
The 8 tools
Jetro exposes eight tools to your agent:
| Tool | What it does |
|---|---|
bw_data | Fetch financial data from paid API sources (fundamentals, prices, filings) |
bw_render | Place elements on the canvas (frames, notes, PDFs, embeds) |
bw_save | Save data locally to your workspace (lists, portfolios, projects) |
bw_query | Run SQL queries against your local DuckDB database |
bw_skill | Load a research skill such as scoring models, DCF logic, or sector frameworks |
bw_template | Load a report template for structured output |
bw_parse | Parse documents like PDFs and DOCX files into structured text |
bw_canvas | Manage canvases. Create, delete, add elements, set up refresh bindings |
Your agent decides when to call which tool based on your request. You just talk to it naturally.
Compatible agents
Jetro works with any MCP-compatible client:
- Claude Code (terminal agent)
- Cursor (built-in agent)
- Windsurf (built-in agent)
- Cline (VS Code extension)
- Any other tool that supports the MCP protocol
Auto-discovery
When Jetro activates, it generates an MCP config file at .berrywise/mcp-config.json. This file tells your agent where to find Jetro’s tools and how to call them. You don’t have to configure anything manually. The agent picks up the config and starts using the tools right away.
You never need to reference these tools by name. Just ask your agent what you want. “Show me Sun Pharma’s revenue trend” or “Build a DCF model for Infosys.” The agent figures out which tools to call on its own.