Integration

Verify Engram from the official MCP Inspector

MCP Inspector is Anthropic's official visual testing tool for MCP servers (often called the MCP Playground). It runs locally via `npx`, opens a browser UI at `http://localhost:6274`, and connects to any MCP server over stdio, SSE, or streamable HTTP. Wire it to Engram's hosted SSE endpoint and you can poke at all six memory tools — store, query, list, list buckets, delete, clear — without writing any client code, which makes it the right surface for debugging auth, schema, and tool behavior before plugging Engram into a real agent.

Install

Three steps: sign up for an Engram API key, paste a BYOK LLM-provider key on /models, then drop the snippet below into MCP Inspector.

Three steps to memory in your agent

  1. Sign up. Free, no card. You'll land on a Getting Started page that walks the next two steps.
  2. Add your LLM key. Engram is BYOK. Paste an OpenAI / Anthropic / Groq / Together / Fireworks key and we'll route every extraction and query call through your provider. You pay your provider directly. We never see your inference.
  3. Paste the snippet below into your agent and restart it. Use Authorization: Bearer <api-key>with the API key from your portal.

Anthropic's official MCP Inspector (a.k.a. "MCP Playground")

Runs locally via npx, opens a browser UI at http://localhost:6274, and connects to any MCP server. The right surface for poking at Engram's six tools before wiring them into a real agent.

  1. Launch the Inspector:
  2. Terminal
    npx -y @modelcontextprotocol/inspector
  3. Open the printed http://localhost:6274/?MCP_PROXY_AUTH_TOKEN=... URL. In the sidebar set Transport to SSE, URL to https://mcp.lumetra.io/mcp/sse, and open the Authentication panel.
  4. Paste your Bearer token (the Inspector's config file intentionally does NOT carry secrets):
  5. Authentication header
    Authorization: Bearer <api-key>
  6. Click Connect, then exercise tools from the Tools tab. For CI / smoke tests, use --cli mode:
  7. Terminal
    npx -y @modelcontextprotocol/inspector \
      --cli https://mcp.lumetra.io/mcp/sse \
      --transport sse \
      --header "Authorization: Bearer <api-key>" \
      --method tools/list

What you can do once memory's wired in

  • Click `Connect` against `https://mcp.lumetra.io/mcp/sse` with your `eng_live_...` key and exercise all six tools from the Tools tab
  • Use the `mcp-config.example.json` + `--server engram` flags to skip the sidebar config every time
  • Headless smoke test in CI: `npx -y @modelcontextprotocol/inspector --cli https://mcp.lumetra.io/mcp/sse --transport sse --header 'Authorization: Bearer eng_live_...' --method tools/list`
  • Debug a 401 / 412 from Engram against the Inspector's raw JSON response before chasing it through an agent stack

FAQ

Is this the same as the 'MCP Playground' people mention?

Yes. `@modelcontextprotocol/inspector` is the canonical npm package; 'MCP Playground' is the colloquial name. The UI runs on `localhost:6274` (client) and `localhost:6277` (proxy) — MCPI / MCPP on a T9 keypad, the project's mnemonic.

Why isn't the bearer key in the config file?

The Inspector's config format intentionally does NOT carry secrets — the `Authorization` header has to be entered in the UI's Authentication panel (or passed via `--header` in `--cli` mode). This is by design: the same config file can be shared across teammates without leaking keys.

Inspector opens but the browser tab is blank / Unauthorized — what now?

Copy the full `http://localhost:6274/?MCP_PROXY_AUTH_TOKEN=...` URL from the terminal; the proxy requires that token on first load. Do NOT set `DANGEROUSLY_OMIT_AUTH=true` to dodge it — that exposes your machine to drive-by browser attacks.

Ship durable memory in MCP Inspector today

Free tier: 10K memories and 50K retrievals per month. No credit card. Same Engram backend powers all 41 integrations, so memories you write from one client are immediately queryable from the rest.