> 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/getting-started/concepts.md).

# Core concepts

The core Algenta model — one integrated engine, governed data, deterministic decisions, capability routing, and explicit deployment boundaries.

Algenta is a governed computational runtime exposed through one versioned HTTP contract. The Python SDK, TypeScript SDK, CLI, and MCP server are clients of that contract. The deployment owns the compiled compute runtime; client users do not install or manage a separate engine component.

## Execution boundary

The public boundary is an Algenta base URL plus an API key:

| Target        | Base URL                     | Who operates compute    |
| ------------- | ---------------------------- | ----------------------- |
| Algenta cloud | `https://api.algenta.ai`     | Algenta                 |
| Self-hosted   | your explicit deployment URL | your Algenta deployment |

Self-hosted clients must receive an explicit private base URL. A private deployment does not silently send work to Algenta cloud. The same `/v1` request and response schemas apply to both targets.

{% hint style="info" %}
The Python package is `algenta-sdk` and imports as `decision_engine`. The TypeScript package is `algenta-sdk`. Both call the integrated Algenta service; neither requires Mojo or worker setup.
{% endhint %}

## Governed data

Data moves through a deliberate lifecycle:

1. **Connect** a database, warehouse, object store, API, file, or repository.
2. **Verify** credentials once and persist the connector state.
3. **Onboard** a table, query, or object as a dataset.
4. **Resolve** structured intent against the known schema.
5. **Query** the validated plan and receive result metadata.

The engine does not guess missing identifiers or execute free-form SQL by default. A governed result carries evidence such as a schema revision, plan hash, validation state, and request id.

## Capability plane

The capability plane catalogs executable resources independently of the client using them.

| Kind              | Meaning                                              |
| ----------------- | ---------------------------------------------------- |
| `dataset`         | A governed data source.                              |
| `skill`           | A reusable instruction or callable behavior.         |
| `mcp_tool`        | A tool exposed by an MCP provider.                   |
| `runtime_library` | A compiled function owned by the Algenta deployment. |

The core operations are discover, route, and execute. Every capability declares an `execution_owner`:

| Owner             | Contract                                                         |
| ----------------- | ---------------------------------------------------------------- |
| `algenta_managed` | Executes inside the selected Algenta deployment.                 |
| `client_managed`  | Executes only through a customer-registered adapter or provider. |

If no valid owner or adapter exists, execution fails explicitly. Algenta does not invent an executor or report a silent success.

## Decisions and simulations

Simulations evaluate a typed scenario with an explicit run count and seed. Decision routes add options, constraints, confidence, risk, and a ranked recommendation. Agent runs add persisted steps, checkpoints, events, and approval gates around longer workflows.

These surfaces share the same controls:

* schema validation before execution;
* authorization and organization scope;
* deterministic compute for the same input, seed, and engine version;
* structured errors instead of hidden fallback;
* request and evidence identifiers for audit.

## Integrated runtime

The control plane handles validation, security, connectors, and persistence. The bundled compiled runtime handles numerical kernels. Algenta keeps workers warm inside the deployment and supervises their lifecycle. A missing or invalid runtime is a deployment failure, not a user setup step and not a reason to switch to Python compute.

Use `GET /v1/infrastructure` with an API key to execute the supported compute probe. A healthy primary node reports `status: "live"` and `engine_mode: "binary"`.

## Privacy profiles

`ALGENTA_DEPLOYMENT_MODE` selects the deployment privacy policy. `self_hosted` and `air_gapped` disable Algenta-cloud execution and apply fail-closed egress controls. Customer connectors and operator services must be explicitly configured and, under strict profiles, allowlisted.

Telemetry and metering modes are independent of compute. A private deployment can retain local audit records without sending them to a control plane.

## Where to go next

{% content-ref url="/pages/oWVTrgfD6zfzaggU9IRO" %}
[Connect a data source](/guides/connect-data.md)
{% endcontent-ref %}

{% content-ref url="/pages/eDDxcUfD56Kbb5TWZoYP" %}
[Run a decision](/guides/run-a-decision.md)
{% endcontent-ref %}

{% content-ref url="/pages/yhNYrVe9zjfJTh9Jj7Pl" %}
[Verify the bundled engine](/engine-runtime/bundled-runtime.md)
{% endcontent-ref %}

{% content-ref url="/pages/pJGCOrvzSg34BTTmhYkb" %}
[Deployment modes & privacy](/concepts/deployment-modes.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/getting-started/concepts.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.