# Modal Stack

In the [AI Enablement Stack](/docs/introduction/ai-enablement-stack), **Blekline is Layer 4** and **Modal is Layer 1** (serverless sandboxes with optional GPU).

| Layer | Product | Role |
|-------|---------|------|
| **L4** | Blekline | Mask prompts, classify risk, govern MCP tool calls, audit |
| **L1** | Modal | gVisor-isolated sandboxes, GPU, high concurrency |

## Wiring

1. Connect `@blekline/mcp-proxy` in Cursor / Claude / Codex — [MCP proxy](/docs/mcp/proxy)
2. Set downstream MCP to a Modal-compatible server (e.g. community `mcp4modal_sandbox` or your Modal app MCP)
3. Provide `MODAL_TOKEN_ID` and `MODAL_TOKEN_SECRET`

```bash
export BLEKLINE_DOWNSTREAM_MCP_COMMAND="npx,-y,mcp4modal_sandbox"
export BLEKLINE_DOWNSTREAM_SERVER=modal
export MODAL_TOKEN_ID=...
export MODAL_TOKEN_SECRET=...
```

**Flow:** L5 Agent → L4 Blekline proxy → L1 Modal MCP → sandbox.

See [Modal Sandboxes documentation](https://modal.com/docs/guide/sandbox) for SDK and MCP setup.

---

**Next steps:** [Sandbox providers](/docs/integrations/sandbox-providers) · [AI Enablement Stack](/docs/introduction/ai-enablement-stack) · [MCP Cursor setup](/docs/mcp/cursor) · [Open workspace](https://app.blekline.com)