Integration: VS Code

Coverage status

Path	Status	Notes
MCP server	✅ Supported	Works with Copilot Chat, Continue.dev, Cline, and any MCP-aware VS Code extension
VS Code extension	✅ Supported (Pro)	Inline context injection; Pro tier only
Harness	—	Not required; MCP is the preferred path

Confidence: High. VS Code’s MCP support has been stable since the 1.96 release. The extension is tested on every MemStrata release.

Setup: MCP path (recommended, any tier)

The MCP path works with any VS Code extension that supports the Model Context Protocol — Copilot Chat, Continue.dev, Cline, and others.

1. Start the MemStrata MCP server

memstrata serve --port 8080

Or add it to your shell profile so it starts on login:

# ~/.zshrc or ~/.bashrc
memstrata serve --port 8080 &

2. Register MemStrata in your VS Code MCP client

For Continue.dev (.continuerc.json or ~/.continue/config.json):

{
  "models": [...],
  "contextProviders": [
    {
      "name": "memstrata",
      "params": {
        "endpoint": "http://localhost:8080/mcp"
      }
    }
  ]
}

For Cline (cline_mcp_settings.json):

{
  "mcpServers": {
    "memstrata": {
      "url": "http://localhost:8080/mcp",
      "disabled": false
    }
  }
}

For Copilot Chat, open the Command Palette → GitHub Copilot: Add MCP Server and enter http://localhost:8080/mcp.

3. Verify

In your AI tool, ask: @memstrata status. You should get back your current context stats and the active project path.

Setup: VS Code extension (Pro tier)

The VS Code extension provides deeper integration: it can inject context into any inline completion request and the sidebar shows your live savings dashboard.

1. Install from the marketplace

Search for “MemStrata” in the VS Code Extensions panel, or:

code --install-extension memstrata.memstrata-vscode

2. Authenticate

The extension will prompt you to log in with your MemStrata license key. This connects the extension to the local harness (no data leaves your machine during auth).

3. Configure

In VS Code Settings (Ctrl+,), search “MemStrata”:

Setting	Default	Notes
`memstrata.harnessPort`	`8080`	Change if 8080 is in use
`memstrata.inlineInjection`	`true`	Injects context into inline completions
`memstrata.dashboardPanel`	`true`	Shows savings sidebar panel

What’s captured

Signal	Captured?	Notes
Copilot Chat messages	✅ Yes	Via MCP or `@memstrata` mention
Inline completions (Copilot, Tabnine)	✅ Yes (Pro extension)	Extension intercepts the completion request
Continue.dev chat	✅ Yes	Via MCP context provider
Cline agent turns	✅ Yes	Via MCP server
GitHub Copilot Workspace	⚠️ Partial	Session-level capture only; per-turn not available
File save events	✅ Yes	Used for 24h edit survival tracking

Troubleshooting

@memstrata status returns “not found”
The MCP server isn’t running. Run memstrata serve and check memstrata status in a terminal.

Extension shows “disconnected”
Check that the harness is running on the configured port (memstrata.harnessPort). Try curl http://localhost:8080/health — it should return {"status":"ok"}.

Inline injection not working
Confirm you’re on Pro tier (memstrata status shows your tier). Inline injection requires the VS Code extension, not just the MCP server.

Port 8080 conflict
Change the port: memstrata serve --port 8181 and update memstrata.harnessPort in VS Code settings accordingly.