Integration

Long-term memory for gptme

gptme is the terminal AI agent: Python, scriptable, and free of UI flash. The Engram plugin is a native gptme plugin. Run `pip install gptme-engram`, set `ENGRAM_API_KEY`, and gptme discovers the six memory tools through its standard entry-point mechanism. There's no MCP bridge and no TOML config to edit.

Install

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

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.

engram-gptme: terminal AI agent

Routes through mcp-remote (stdio→SSE bridge) since Engram's MCP endpoint is SSE-only and gptme defaults to Streamable HTTP. Source: github.com/lumetra-io/engram-gptme.

  1. Export your API key:
  2. Terminal
    export ENGRAM_API_KEY="<api-key>"
  3. Add this block to ~/.config/gptme/config.toml:
  4. ~/.config/gptme/config.toml
    [mcp]
    enabled = true
    
    [[mcp.servers]]
    name = "engram"
    enabled = true
    command = "npx"
    args = ["-y", "mcp-remote", "https://mcp.lumetra.io/mcp/sse", "--transport", "sse-only", "--header", "Authorization:Bearer ${ENGRAM_API_KEY}"]
  5. Run gptme --tools mcp "list available MCP tools and exit". The six Engram tools should appear in the agent's tool palette.

What you can do once memory's wired in

  • Add a memory layer to gptme without touching its core config
  • Pin a default bucket per project via `gptme-engram`'s env-var config so each repo gets its own memory
  • Chain gptme tasks across days, so the agent remembers what the previous run accomplished
  • Self-host Engram and point gptme at it via `ENGRAM_API_BASE` for a fully local stack

FAQ

Is this still the `mcp-remote` bridge setup from before?

No. `gptme-engram` is a native plugin now: pure pip install, with no `npx mcp-remote` and no TOML edits. The MCP-bridge approach is deprecated for gptme specifically.

Where does it look for the API key?

`ENGRAM_API_KEY` in the environment is the canonical source. If you installed gptme via pipx, use `pipx inject gptme gptme-engram` so the plugin lives in the same environment as the CLI.

Can I override the API base URL?

Yes. `ENGRAM_API_BASE=https://engram.internal.example.com` points the plugin at a self-hosted Engram instance. Same six tools, same agent surface, different backend.

Ship durable memory in gptme 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.