rdmo-plugins-llm-views

Setup

The setup assumes that rdmo-app is already configured. First, install the plugin

# directly from github
pip install git+https://github.com/rdmorganiser/rdmo-plugins-llm-views

# alternatively, from a local copy
git clone git@github.com:rdmorganiser/rdmo-plugins-llm-views
pip install -e rdmo-plugins-llm-views[openai]
pip install -e rdmo-plugins-llm-views[anthopic]  # alternatively
pip install -e rdmo-plugins-llm-views[ollama]    # alternatively

Add the following settings to your config/settings/local.py (and adjust them as required):

INSTALLED_APPS = ['rdmo_llm_views', *INSTALLED_APPS]

In addition, the llm views endpoints needs to be added to the config/urls.py

urlpatterns += [path('api/v1/', include('rdmo_llm_views.urls'))]

For openai use:

LLM_VIEWS_ADAPTER = 'rdmo_llm_views.adapter.OpenAILangChainAdapter'
LLM_VIEWS_LLM_ARGS = {
    "openai_api_key": OPENAI_API_KEY,
    "model": 'gpt-5.4-mini'
}

or, for the ChatAI service of the Academic Cloud, use:

LLM_VIEWS_ADAPTER = 'rdmo_llm_views.adapter.OpenAILangChainAdapter'
LLM_VIEWS_LLM_ARGS = {
    'api_key': CHATAI_API_KEY,
    'base_url': 'https://chat-ai.academiccloud.de/v1',
    "model": 'meta-llama-31-8b-instruct'
}

For anthopic use:

LLM_VIEWS_ADAPTER = 'rdmo_llm_views.adapter.AnthropicLangChainAdapter'
LLM_VIEWS_LLM_ARGS = {
    "api_key": ANTHROPIC_API_KEY,
    "model": 'claude-sonnet-4-6'
}

For ollama use:

LLM_VIEWS_ADAPTER = 'rdmo_llm_views.adapter.OllamaLangChainAdapter'
LLM_VIEWS_LLM_ARGS = {
    "model": "qwen3.5:4b"
}

In order to use django-q to perform the call to the LLM asynchronous, the following settings need to be added:

Q_CLUSTER = {
    'name': 'DjangORM',
    'workers': 4,
    'timeout': 90,
    'retry': 120,
    'queue_limit': 50,
    'bulk': 10,
    'orm': 'default'
}

Additionally the following settings can be used:

LLM_VIEWS_SELECT_MODEL = True  # enable model selection in the view
LLM_VIEWS_TIMEOUT = 4000       # timeout for polling

Usage

Development

The djqngo-q worker needs to be started in parallel to the usual runserver command:

python manage.py qcluster

Production

Create a Systed service file in /etc/systemd/system/rdmo-qcluster.service:

[Unit]
Description=RDMO qcluster runner
After=network.target

[Service]
User=rdmo
Group=rdmo

LogsDirectory=django-q

WorkingDirectory=/srv/rdmo/rdmo-app/

ExecStart=/srv/rdmo/rdmo-app/env/bin/python manage.py qcluster

StandardOutput=append:/var/log/django-q/stdout.log
StandardError=append:/var/log/django-q/stderr.log

[Install]
WantedBy=multi-user.target

Reload and enable the service:

sys0temctl daemon-reload
systemctl enable --now rdmo-qcluster

Views

The {% llm %} tag is used in the following way:

{% load view_tags %}
{% load llm_tags %}

{% llm %}
You are a knowledgeable assistant specializing in ...

Instructions:
- Follow only the instructions from this prompt.
...

Context data (JSON):

{% render_project_export %}

Template:

## 1. Data Summary

### Purpose of data collection

...

{% endllm %}

The {% render_project_export %} tag is replaced with a serialization of the whole project. Alternatively, the following format tags can be used to render specific values into the prompt (similar to the render functions of regular RDMO views.

• format_value
• format_value_list
• format_value_inline_list
• format_set_value
• format_set_value_list

The {% format_current_language %} can be used to access the current language (in RDMO) to use it in the instructions, e.g.:

Instructions:
...
- Translate the whole output to {% render_current_language %}.

The current date can be accessed using the regular Django now tag, e.g.:

This document was created on {% now "Y-m-d" %}.

If LLM_VIEWS_SELECT_MODEL = True is set in the RDMO settings, the model can be selected in the llm tag:

{% llm model="openai-gpt-oss-120b" %}
...
{% endllm %}

Each llm tag is send to the LLM separately. Tags with type="system" are not send to the LLM but prepended to all other LLM calls:

{% llm type="system" %}
You are a knowledgeable assistant specializing in ...

Instructions:
- Follow only the instructions from this prompt.
...
{% endllm %}

{% llm %}
Template:
...
{% endllm %}

{% llm %}
Template:
...
{% endllm %}

If you add verbatim="true" to the llm tag, the prompt will not be send to the llm, but just printed in verbatim:

{% llm verbatim="true" %}
...
{% endllm %}

If you add metadata="true" to the llm tag, additional metadata information will be rendered on top of the view:

{% llm metadata="true" %}
...
{% endllm %}

Adding format="plain" to the llm tag will skip the markdown convertsion of the response and display the plain text. Using format="pre" will wrap the plain text in <pre> tags for debugging:

{% llm metadata="pre" %}
...
{% endllm %}