mcp-agent-to-vsc
# MCP - Agent to VSC
Local MCP server that connects an MCP-capable VS Code agent such as Cline to OpenAI and workspace file tools.
## Flow
`Cline / VS Code -> MCP (stdio) -> local Node.js server -> OpenAI API`
## Tools
- `ask_ai` — send prompts to OpenAI
- `read_file` — read workspace files
- `write_file` — write workspace files
- `list_files` — list workspace files/directories
- `workspace_info` — inspect configuration
## Install
Requirements: Node.js 20+.
```bash
npm install
```
Set `OPENAI_API_KEY` in the environment. Never commit the real key.
PowerShell:
```powershell
$env:OPENAI_API_KEY="sk-..."
npm start
```
## Cline
Add this server to Cline's MCP settings:
```json
{
"mcpServers": {
"mcp-agent-to-vsc": {
"command": "node",
"args": ["C:\\path\\to\\mcp-agent-to-vsc\\server.mjs"],
"env": {
"OPENAI_API_KEY": "YOUR_API_KEY",
"OPENAI_MODEL": "gpt-5.6",
"WORKSPACE_ROOT": "C:\\path\\to\\your\\workspace"
}
}
}
}
```
For local VS Code MCP configuration, see `.vscode/mcp.json`.
## Security
The server restricts file access to `WORKSPACE_ROOT`. Keep your API key in an environment variable or an ignored `.env` file.
TDQS
Scored across 5 tools
Each tool targets a distinct function: AI query, file read, file write, file listing, and workspace configuration. There is no functional overlap, so an agent can reliably select the correct tool.
Most names follow a verb_noun pattern (ask_ai, read_file, write_file, list_files), but workspace_info breaks the pattern as a noun phrase. Overall the naming is clear and predictable.
Five tools is a well-scoped set for an agent-to-VS Code integration covering the essential file operations, AI interaction, and workspace context. Each tool earns its place without redundancy.
The set covers core file read/write/list and AI query, which covers common agent workflows. Minor gaps include delete or rename file operations, but these are not critical for the apparent purpose.