> For the complete documentation index, see [llms.txt](https://docs.algenta.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.algenta.ai/run-the-engine/filtered-install.md).

# Filtered & selective install

Algenta is a large system, but you almost never want all of it. Install **only what your use case needs** — a thin client, a local runtime, or the full engine — and, when you do run the engine, compile **only the modules you use** (a full build is \~17 GB). This page is the map of every "less than everything" choice.

## 1. Pick a component profile

| Profile              | Install                                                  | What you get                                                                                                                            | Use when                                                                      |
| -------------------- | -------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
| **SDK only**         | `pip install algenta-sdk` (or `npm install algenta-sdk`) | A typed HTTP client for a running API. No engine, no local compute.                                                                     | You call a hosted or self-hosted Algenta from your app.                       |
| **Local runtime**    | `pip install algenta-core`                               | The in-process `Runtime` for local resolve/query flows. Add `[cloud]` for the API transport, `[benchmark]` for numpy, `[all]` for both. | You run deterministic flows locally or drive a remote engine from one object. |
| **MCP server**       | `pipx install algenta-mcp`                               | The MCP bridge for Cursor / Claude Desktop. Ships independently.                                                                        | You give an AI client engine tools.                                           |
| **Full self-hosted** | `curl -sSL https://algenta.ai/install.sh \| bash`        | API + worker + **Mojo engine** + Postgres + Redis + object storage (Docker Compose).                                                    | You run the complete decision engine on your own infra.                       |

{% hint style="info" %}
The extras are additive: `pip install "algenta-core[cloud]"` is the local runtime **plus** the HTTP transport. Only the full self-hosted profile pulls the Mojo engine and the data services.
{% endhint %}

## 2. Compile only the engine modules you need

When you build the engine from source, **`MOJO_RUNTIME_MODULE_FILTER`** selects which runtime-library modules are compiled into the binary. An empty filter builds everything (\~17 GB of work); a filter builds a small, fast binary with just your modules.

```bash
# build a lean binary with only the modules you actually call
MOJO_RUNTIME_MODULE_FILTER="monte_carlo,distributions,optimization_constrained" make build-mojo
```

Confirm what made it in:

```bash
curl http://localhost:8000/v1/runtime/manifest    # lists the compiled modules
```

{% hint style="warning" %}
A request for a module that wasn't compiled in fails with **"module not registered"**. If you filter, include every module your workloads use — or rebuild without the filter for the full set.
{% endhint %}

## 3. Expose only one edition

A single engine can present a curated **edition** surface (Codna, Cohenta, …) instead of every tool. Set **`ALGENTA_PRODUCT`** (or send `X-Algenta-Product` per request); the MCP/API tool set is filtered to that edition's allowlist (unknown product → empty set, fail-closed).

```bash
ALGENTA_PRODUCT=codna     # only this edition's tools are visible
```

## 4. Run only the control plane

A deployment runs in one of two roles via **`ALGENTA_SERVICE_ROLE`**:

| Role             | Includes                                                              | Excludes                   | Use for                                                       |
| ---------------- | --------------------------------------------------------------------- | -------------------------- | ------------------------------------------------------------- |
| `full` (default) | Everything — engine, simulate, agent runs, `/mcp`, the Mojo pool      | —                          | The complete service.                                         |
| `control_plane`  | Auth, accounts, teams, billing, credits, metering, devices, licensing | The engine + Mojo entirely | A tiny, engine-free auth/billing tier that scales separately. |

The runtime-library HTTP endpoints (`/v1/libraries*`) are **off by default**; enable them only if you need direct module invocation:

```bash
ALGENTA_SERVICE_ROLE=control_plane            # engine-free control plane
ALGENTA_RUNTIME_LIBRARY_HTTP_API_ENABLED=1    # expose /v1/libraries* (off by default)
```

{% hint style="success" %}
**Expected result** — you install (and build) only the surface your use case needs: a 5 MB SDK, a local runtime, an engine-free control plane, or a lean engine with just your modules — instead of the full stack.
{% endhint %}

## Next

{% content-ref url="/pages/zckshsBx1v9QwYvImRX6" %}
[Run the Mojo engine](/run-the-engine/mojo-sidecar.md)
{% endcontent-ref %}

{% content-ref url="/pages/tWtESdNlxR5ZqT7F3czC" %}
[Self-hosting](/deploy-and-operate/self-hosting.md)
{% endcontent-ref %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.algenta.ai/run-the-engine/filtered-install.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.