> For the complete documentation index, see [llms.txt](https://robodex.gitbook.io/robodex-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://robodex.gitbook.io/robodex-docs/product-api/workspace-sdk.md).

# Workspace SDK

`apps/api/workspace_sdk.py` contains `RobotWorkspaceSDK`, the backend-first orchestration surface for the workspace UI.

{% hint style="info" %}
Keep frontend and Electron code thin. If a workflow changes project, thread, generation, simulation, or policy state, it belongs in the backend SDK or store.
{% endhint %}

## Core responsibilities

`RobotWorkspaceSDK` owns:

* project creation and listing,
* thread creation and listing,
* user and assistant message persistence,
* prompt generation orchestration,
* context summary creation,
* selected design tracking,
* simulation spec creation,
* lightweight simulation checks,
* policy spec creation.

## Generation flow

`generate_for_thread()` is the main entrypoint.

```mermaid
flowchart TD
    A["Validate prompt"] --> B["Load thread and project"]
    B --> C["Choose agent loop"]
    C --> D["Append user message"]
    D --> E["Create synthetic ingest job"]
    E --> F["Call design generation route"]
    F --> G["Attach robot_design artifact"]
    G --> H["Set selected design"]
    H --> I["Append assistant message"]
    I --> J["Create context summary"]
```

## Loop selection

The SDK chooses a loop conservatively:

1. use the requested loop if it is registered,
2. otherwise prefer `creative_qd_v2` if available,
3. otherwise fall back to `grammar_v2`.

This keeps the product path usable even if a more experimental loop is unavailable in a local checkout.

## Synthetic ingest jobs

Workspace generation does not require the user to run `/ingest` first. The SDK creates an ingest job with:

* the prompt,
* a minimal plan,
* source `none`,
* status suitable for generation.

It then calls the same design generation route used by ingest-driven flows. This avoids duplicating candidate creation logic.

## Thread artifacts

Generated designs are saved as thread artifacts with type `robot_design`. The artifact contains:

* selected candidate,
* all returned candidates,
* render payloads,
* telemetry,
* model preference,
* grammar HITL data.

Frontend panels read these artifacts rather than recomputing generation state.

## Simulation checks

`run_simulation_checks()` is intentionally lightweight. It checks whether:

* MJCF or render payload data exists,
* the candidate is marked compile-safe,
* screening scores clear the local threshold when available.

It records `full_training_started: False`. Full RL training is not hidden behind this route.

## Policy specs

`create_policy_spec()` stores a policy-training request with observation, action, reward, and safety fields. It marks the spec ready but does not run PPO or any long-running control training.


---

# 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://robodex.gitbook.io/robodex-docs/product-api/workspace-sdk.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.