Content
# MCP Projects
Documenting my MCP (Model Context Protocol) server connections as I learn AI tooling — first step toward AI, quant finance, and eventually trading systems.
## What is MCP?
MCP (Model Context Protocol) is an open standard that lets AI assistants like Claude connect to external tools and data sources — calendars, file systems, databases, note-taking apps, etc. Each connection is called a "server," and it exposes a set of actions the AI can call.
## Connections
### Obsidian (via mcp-obsidian)
Connects Claude Desktop to a local Obsidian vault, enabling the AI to read, search, and create notes directly.
**Stack:**
- [`mcp-obsidian`](https://github.com/MarkusPfundstein/mcp-obsidian) — Python MCP server (run via `uvx`, no manual install needed)
- Obsidian community plugin: **Local REST API with MCP** — exposes the vault over a local HTTPS API
- `uv` / `uvx` — Python package runner
**How it works:**
1. Obsidian runs the Local REST API plugin, serving the vault on `127.0.0.1:27124`
2. `mcp-obsidian` bridges that REST API to the MCP protocol
3. Claude Desktop launches the bridge automatically via config and can then read/write notes
**Config shape** (`claude_desktop_config.json`):
\`\`\`json
{
"mcpServers": {
"mcp-obsidian": {
"command": "uvx",
"args": ["mcp-obsidian"],
"env": {
"OBSIDIAN_API_KEY": "YOUR_KEY_HERE",
"OBSIDIAN_HOST": "127.0.0.1",
"OBSIDIAN_PORT": "27124"
}
}
}
}
\`\`\`
*(API key intentionally omitted — never commit real keys to a public repo.)*
**What it enables:** the AI can organize notes, build study trackers, and maintain a structured knowledge base directly inside my vault.
## Roadmap
- [ ] Add more MCP connections as I set them up
- [ ] Python + quant tooling notes
- [ ] Eventually: AI-assisted trading research tools
---
*Student at Saint Paul College, math & CS, building toward quant finance.*
