Screenshots
AI Preprocessing Assistant
PROAI Preprocessing Assistant module for Zabbix Module to configure OpenAI, Google Gemini, Anthropic Claude or any OpenAI-compatible endpoint together with the model, API key, system role and initial prompt. Configuration options include provider selection, model choice, stored API key and editable system/initial prompts; all provider calls are performed by the Zabbix backend so the API key never reaches the browser. This module assists Zabbix administrators in generating, reviewing and writing preprocessing steps and the item output type directly into the item form.
Zabbix 6.0 - 8.0
How to install
Step 1 · Operating system
Step 3 · Installation
PRO repositories are protected with HTTP basic auth. You need one personal repo token, then you point your package manager at the authenticated URL. A single token unlocks every product your account is entitled to - you do not need a token per widget.
wget https://repo.initmax.com/zabbix/release/initmax-release-latest_all.deb
dpkg -i initmax-release-latest_all.deb
printf 'machine repo.initmax.com login token password %s\n' 'TOKEN_VALUE' > /etc/apt/auth.conf.d/initmax.conf
chmod 600 /etc/apt/auth.conf.d/initmax.conf
sed -i -e '/^# Types: deb/,$ s/^# //' -e 's|USER:TOKEN@||' /etc/apt/sources.list.d/initmax.sources
apt update && apt install zabbix-module-ai-preprocessing-assistantrpm -Uvh https://repo.initmax.com/zabbix/release/initmax-release-latest.noarch.rpm
sed -i -e '/#\[initmax-zabbix-pro\]/,$ s/^#//' -e 's|USER:TOKEN|token:TOKEN|' /etc/yum.repos.d/initmax.repo
chmod 600 /etc/yum.repos.d/initmax.repo
dnf clean all && dnf install -y zabbix-module-ai-preprocessing-assistantrpm -Uvh https://repo.initmax.com/zabbix/release/initmax-release-latest.noarch.rpm
sed -i -e '/#\[initmax-zabbix-pro\]/,$ s/^#//' -e 's|USER:TOKEN|token:TOKEN|' /etc/zypp/repos.d/initmax.repo
chmod 600 /etc/zypp/repos.d/initmax.repo
zypper --gpg-auto-import-keys refresh && zypper install -y zabbix-module-ai-preprocessing-assistantOne token unlocks every PRO product on your account. If you cancel one product, the same token keeps working for the rest. Rotate or revoke tokens any time from the Portal.
Step 4 · Edit your Dockerfile
Paste the block after FROM in the final Zabbix frontend stage.
# initMAX ai-preprocessing-assistant - Alpine (apk) package install
USER root
RUN --mount=type=secret,id=initmax_repo_token,required=true \
apk add --no-cache ca-certificates curl \
&& curl -fsSL https://repo.initmax.com/zabbix/release/setup-alpine.sh -o /tmp/initmax-repo.sh \
&& sh /tmp/initmax-repo.sh --edition pro \
--token-file /run/secrets/initmax_repo_token \
-- apk add --no-cache zabbix-module-ai-preprocessing-assistant \
&& rm -f /tmp/initmax-repo.sh
USER 1997On the last line, keep the USER from your original image.
The block installs the newest available version when the installation step runs. Docker may reuse a cached layer. To keep a fixed version, use zabbix-module-ai-preprocessing-assistant=2.3.4_p2-r3
Step 5 · Build your image
Reveal or create a repo token above. Run the filled command in a terminal in the Dockerfile directory; do not save it to a file.
# initMAX ai-preprocessing-assistant - Alpine (apk) package install
INITMAX_REPO_TOKEN='TOKEN_VALUE' \
docker build --secret id=initmax_repo_token,env=INITMAX_REPO_TOKEN \
-t zabbix-web-custom .In Compose, use image zabbix-web-custom and recreate the frontend. In Zabbix, open Administration / General / Modules, Scan directory and enable the module.
More options
PRO uses your existing repo token as a Docker build secret. Do not put the token in Dockerfile instructions or ARG/ENV.
Buying another product later? Add its install block to the same Dockerfile; keep the existing blocks and products.
Step 4 · Edit your Dockerfile
For Ubuntu / Debian frontend images. Add after FROM. Keep installed modules in the image; a volume over modules/ would hide package updates.
USER root
RUN --mount=type=secret,id=initmax_repo_token,required=true \
apt-get update \
&& apt-get install -y --no-install-recommends ca-certificates curl \
&& curl -fsSL https://repo.initmax.com/zabbix/release/initmax-release-latest_all.deb -o /tmp/initmax-release.deb \
&& dpkg -i /tmp/initmax-release.deb \
&& cp /etc/apt/sources.list.d/initmax.sources /tmp/initmax.sources \
&& printf 'machine repo.initmax.com login token password %s\n' "$(cat /run/secrets/initmax_repo_token)" > /etc/apt/auth.conf.d/initmax.conf \
&& chmod 600 /etc/apt/auth.conf.d/initmax.conf \
&& sed -i -e '/^# Types: deb/,$ s/^# //' /etc/apt/sources.list.d/initmax.sources \
&& apt-get update \
&& apt-get install -y --no-install-recommends zabbix-module-ai-preprocessing-assistant \
&& mv /tmp/initmax.sources /etc/apt/sources.list.d/initmax.sources \
&& rm -f /tmp/initmax-release.deb /etc/apt/auth.conf.d/initmax.conf \
&& rm -rf /var/lib/apt/lists/*
USER 1997On the last line, keep the USER from your original image.
Step 5 · Build your image
Reveal or create a repo token above. Run the filled command in a terminal in the Dockerfile directory; do not save it to a file.
INITMAX_REPO_TOKEN='TOKEN_VALUE' \
docker build --secret id=initmax_repo_token,env=INITMAX_REPO_TOKEN \
-t zabbix-web-custom .In Compose, use image zabbix-web-custom and recreate the frontend. In Zabbix, open Administration / General / Modules, Scan directory and enable the module.
PRO uses your existing repo token as a Docker build secret. Do not put the token in Dockerfile instructions or ARG/ENV.
Step 4 · Edit your Dockerfile
For CentOS Stream / Oracle Linux frontend images. Add after FROM. DNF installs the RPM package; keep installed modules in the image, not a volume over modules/.
USER root
RUN --mount=type=secret,id=initmax_repo_token,required=true \
if ! command -v dnf >/dev/null; then microdnf install -y dnf; fi \
&& { rpm -q initmax-release >/dev/null 2>&1 || rpm -Uvh https://repo.initmax.com/zabbix/release/initmax-release-latest.noarch.rpm; } \
&& cp /etc/yum.repos.d/initmax.repo /tmp/initmax.repo \
&& sed -i -e '/#\[initmax-zabbix-pro\]/,$ s/^#//' -e 's|USER:TOKEN@||' /etc/yum.repos.d/initmax.repo \
&& printf '\nusername=token\npassword=$initmax_repo_token\n' >> /etc/yum.repos.d/initmax.repo \
&& DNF_VAR_initmax_repo_token="$(cat /run/secrets/initmax_repo_token)" dnf install -y zabbix-module-ai-preprocessing-assistant \
&& mv /tmp/initmax.repo /etc/yum.repos.d/initmax.repo \
&& dnf clean all
USER 1997On the last line, keep the USER from your original image.
Step 5 · Build your image
Reveal or create a repo token above. Run the filled command in a terminal in the Dockerfile directory; do not save it to a file.
INITMAX_REPO_TOKEN='TOKEN_VALUE' \
docker build --secret id=initmax_repo_token,env=INITMAX_REPO_TOKEN \
-t zabbix-web-custom .In Compose, use image zabbix-web-custom and recreate the frontend. In Zabbix, open Administration / General / Modules, Scan directory and enable the module.
PRO uses your existing repo token as a Docker build secret. Do not put the token in Dockerfile instructions or ARG/ENV.
ZIP modules in a persistent volume. You manage updates manually.
For manual PRO installation, sign in with an account holding an active licence.
Have thoughts about AI Preprocessing Assistant? Please send us feedback.
Releases
272.3.4 Latest 2026-11-13
Fix
-
Added Google Gemini model Gemini 4 Hydrogen (
gemini-4-hydrogen).
2.3.3 2026-11-12
Fix
-
Added Google Gemini model Gemini Decision 1 (
gemini-decision-1).
2.3.2 2026-11-11
Fix
-
Added Google Gemini model Gemini 4 Carbon (
gemini-4-carbon).
Older versions (24)
2.3.1 2026-11-10
Fix
-
Added Anthropic Claude model Claude Fable 5.2 (
claude-fable-5-2).
2.3.0 2026-11-05
Feature
-
Publish the Alpine (apk) package for Docker based Zabbix frontends (repo.initmax.com Alpine channel).
2.2.1 2026-10-27
Fix
-
Added OpenAI model GPT-6.1 Terra (
gpt-6.1-terra).
2.2.0 2026-10-15
Feature
-
Support Zabbix 8.0 with updated item handling and settings.
2.1.12 2026-10-14
Fix
-
Added Google Gemini model Gemini 4 Argon (
gemini-4-argon).
2.1.11 2026-10-12
Fix
-
Added Google Gemini model Gemini Nano Banana 2.1 (
gemini-nano-banana-2.1). -
Added Anthropic Claude model Claude Haiku 5.5 (
claude-haiku-5-5).
2.1.10 2026-09-30
Fix
-
Added OpenAI model GPT-6.1 Sol (
gpt-6.1-sol).
2.1.9 2026-09-29
Fix
-
Sorted supported AI models by generation, newest first.
2.1.8 2026-09-29
Fix
-
Added OpenAI model GPT-6 Luna (
gpt-6-luna). -
Added OpenAI model GPT-6 Sol (
gpt-6-sol). -
Added Anthropic Claude model Claude Opus 5.5 (
claude-opus-5-5). -
Added Anthropic Claude model Claude Sonnet 5.5 (
claude-sonnet-5-5).
2.1.7 2026-09-10
-
Fix blank Item popup opened from dashboards
2.1.6 2026-09-09
-
Preserve AI provider credentials while editing
2.1.5 2026-09-09
-
Publish current supported AI models
2.1.4 2026-09-09
-
Sync supported AI models
2.1.3 2026-09-02
-
Correct AI Preprocessing catalog description
2.1.2 2026-08-21
-
the Stream response toggle shows for every provider
2.1.1 2026-08-20
-
the release capture runs again and the fallback screens are current
2.1.0 2026-08-20
-
Server-side providers and the APDA-style administration
2.0.6 2026-08-20
-
en_GB menu label casing AI General, endpoint example as a full URL
2.0.5 2026-08-20
-
Dark themes: the page logo follows the theme
2.0.4 2026-08-17
-
Settings form: help hint on every field
2.0.3 2026-08-17
-
One 'AI General' submenu shared with the other AI products
2.0.2 2026-08-16
-
README on the combo skeleton: portal installer, feature table, same sections everywhere
2.0.1 2026-08-16
-
Catalog showcase spec: a scripted preprocessing conversation
2.0.0 2026-08-10
-
Graft the AI Preprocessing Assistant onto the v3 pipeline