Tasker → n8n → HA → AI: Master Training Curriculum

Part 1

Operational Syllabus & Roadmap

ℹ️Foundation Rule: Each tier unlocks the next. Don't skip. Tiers 1–2 work entirely on your LAN — no cloud needed.

Tasker HTTP Foundations

Make the phone talk JSON to anything

🏆 Win Condition

Tasker POSTs JSON to a test endpoint, reads a response field, shows it in a notification. Error branch fires on bad response code.

Profile → Task → Variable architecture
HTTP Request action (GET / POST)
Headers: Content-Type:application/json
Body: raw JSON with Tasker vars
Native JSON: %http_data.fieldname
Read %http_response_code
Global vs local variables (case rule)
Continue After Error + branch logic

Home Assistant Direct Control

Tasker ↔ HA with no middleman

🏆 Win Condition

Tasker turns a light on/off via HA REST API. HA broadcasts an intent back. Tasker catches it and speaks the result.

Create HA Long-Lived Access Token
GET /api/ — health check
GET /api/states/light.x
POST /api/services/light/turn_on
HA Webhook trigger (local_only: true)
command_broadcast_intent → Tasker
Intent Received profile in Tasker
Intent extras as local variables

n8n as the Hub

Route, transform, orchestrate

🏆 Win Condition

Tasker hits n8n Production webhook → n8n calls HA → n8n returns structured JSON to Tasker. Round trip under 5 seconds.

⚠️Critical gotcha: Test URL only works while n8n editor is listening. Production URL only works when workflow is ACTIVE. Always use Production URL in Tasker.

n8n Docker self-host basics
Webhook node — Header Auth setup
Test vs Production URL (burn this in)
HTTP Request node → HA service call
Respond to Webhook node
Built-in HA node (Service / State ops)
Switch node — intent routing
Accessing {{$json.body.field}}

AI Integration

Gemini, Ollama, structured outputs

🏆 Win Condition

Natural language command on phone → AI parses intent → correct HA service fires → AI-written reply returns to Tasker.

Gemini API direct from Tasker HTTP Request
Store key in global %GEMINI_KEY
Ollama via n8n (Docker networking)
stream: false — always required
JSON Schema output from LLM
n8n AI Agent + HA node as tool
Entity allowlist (security)
/api/conversation/process shortcut

End-to-End Pipelines

Jarvis-grade loops with return paths

🏆 Win Condition

Full async loop: voice trigger → n8n agent → HA executes → AI reply returns as actionable notification with Yes/No buttons. Works on LAN and via Tailscale off-LAN.

command_broadcast_intent — async return
AutoNotification action buttons → webhook
Tasker HTTP server inbound (port 1821)
n8n → Tasker push over LAN
Tailscale for off-LAN access
Full error handling stack
ZeroClaw-style watchdog patterns
Capstone: voice → AI → HA → spoken reply

Part 2

Hands-On Lab Blueprints

Lab 1 · Tier 2

📍 Phone Check-In Reporter

Tasker → HA REST API. Phone sends battery + WiFi status to HA. HA broadcasts confirmation back to phone.

🎯 Objective

No AI. No n8n. Just Tasker ↔ HA REST. Proves the foundation before adding complexity.

Triggers, Variables & Actions

Item	Value
Trigger	Time → Every 30 minutes
`%HA_HOST`	`http://192.168.x.x:8123`
`%HA_TOKEN`	Long-Lived Access Token (global)
`%BATT`	Built-in Tasker var — battery %
`%WIFII`	Built-in Tasker var — WiFi SSID
Action 1	HTTP Request → POST to HA webhook
Action 2	If `%http_response_code` = 200 → Flash "✅ Sent"
Action 3	Else → Flash "❌ %http_response_code"

Tasker HTTP Request

Method: POST URL: %HA_HOST/api/webhook/tasker-checkin-abc123 Headers: Content-Type:application/json Body: { "device": "%DEVICEID", "battery": %BATT, "wifi": "%WIFII", "timestamp": "%DATE %TIME" } Timeout: 10 Continue After Error: ON ✓

HA automation.yaml — webhook → broadcast back

alias: Tasker Check-In Handler triggers: - trigger: webhook webhook_id: tasker-checkin-abc123 allowed_methods: [POST] local_only: true actions: - action: notify.mobile_app_YOUR_PHONE data: message: command_broadcast_intent data: intent_package_name: net.dinglisch.android.taskerm intent_action: com.example.HA_CHECKIN_ACK intent_extras: "battery:{{ trigger.json.battery }}:int"

Tasker Intent Received Profile

// Profile: Event → System → Intent Received Action Filter: com.example.HA_CHECKIN_ACK // Task: Flash: "HA confirmed. Battery: %battery"

✅ 3-Step Verification

1 Run Task manually → Tasker log → %http_response_code = 200

2 HA Developer Tools → Events → fire webhook manually → automation triggers

3 Phone Flash shows battery number echoed back from HA

Lab 2 · Tier 3 + 4

🗣️ AI Light Commander

Voice text → Tasker → n8n → Gemini → HA service call → AI reply spoken on phone. Synchronous round-trip.

🎯 Objective

Speak a command. Gemini parses intent. HA turns lights on/off. AI reply returns in same HTTP response.

📱 Tasker

→

n8n Webhook

→

Gemini API

→

HA Service

→

📱 Reply

Triggers, Variables & Actions

Item	Value
Trigger	Get Voice action → store in `%prompt`
`%N8N_URL`	`http://n8n.local:5678/webhook/android-ai`
`%N8N_SECRET`	Header Auth secret (global)
`%prompt`	Voice input text
`%http_data.reply`	AI reply text from n8n response
`%http_data.action`	HA service that fired
Action 1	Get Voice → `%prompt`
Action 2	HTTP Request → POST to n8n
Action 3	Say → `%http_data.reply`

Tasker HTTP Request — JSON Payload

Method: POST URL: %N8N_URL Headers: Content-Type:application/json X-Webhook-Secret:%N8N_SECRET Body: { "device": "%DEVICEID", "text": "%prompt", "battery": %BATT } Timeout: 30 Continue After Error: ON ✓

n8n Workflow — Key Node Configs

// 1. Webhook node httpMethod: POST path: android-ai authentication: headerAuth responseMode: responseNode ← use Respond to Webhook node // 2. HTTP Request → Gemini Method: POST URL: https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent Header: x-goog-api-key = {{ $env.GEMINI_KEY }} Body: { "contents": [{ "parts": [{ "text": "={{ 'Parse this smart home command. Return ONLY JSON: {\"service\":\"light.turn_on or light.turn_off\",\"entity_id\":\"light.kitchen or light.living_room\",\"reply\":\"friendly 1-sentence reply\"}. Command: ' + $json.body.text }}" }] }] } // 3. Home Assistant node — call the service Operation: Call Service Domain: {{ $json.service.split('.')[0] }} Service: {{ $json.service.split('.')[1] }} Data: { "entity_id": "{{ $json.entity_id }}" } // 4. Respond to Webhook node respondWith: json body: { "ok": true, "reply": "={{ $('Gemini').item.json.reply }}", "action": "={{ $('Gemini').item.json.service }}" }

Tasker — Read Reply & Speak

If %http_response_code = 200 Say: %http_data.reply Else Flash: "Error: %http_response_code" Set Global: %LAST_ERROR = "%http_response_code %http_data" End If

✅ 3-Step Verification

1 n8n test: send POST with {"text":"turn off kitchen lights"} → Gemini returns valid JSON with service and entity_id

2 Activate workflow → Tasker voice test → HA Developer Tools → States → target light changed

3 Phone speaks the AI-generated reply field (not a template string)

Lab 3 · Tier 5

🚨 Smart Alert Processor

HA sensor triggers → n8n → AI writes summary → HA broadcasts to phone → Tasker shows notification with Acknowledge / Dismiss buttons.

🎯 Objective

HA detects something (door left open, temp threshold, low battery sensor). n8n fetches context, AI writes a human-friendly summary. Phone gets notification with action buttons. Button tap fires follow-up webhook.

HA Sensor

→

n8n Webhook

→

Ollama / Gemini

→

HA Notify

→

📱 Tasker Intent

Triggers, Variables & Actions

Item	Value
HA Trigger	State change on `binary_sensor.front_door` → on for 5 min
HA Action	`rest_command.n8n_alert` → POST to n8n
n8n vars	`{{$json.body.entity_id}}`, `{{$json.body.state}}`, `{{$json.body.duration}}`
LLM prompt	System: "Write 1-sentence smart home alert." + entity context
`%summary`	AI text received via intent extra
Intent action	`com.example.HA_ALERT_ACK`

HA configuration.yaml — rest_command

rest_command: n8n_alert: url: http://n8n.local:5678/webhook/ha-alert method: POST headers: X-Webhook-Secret: YOUR_SECRET Content-Type: application/json payload: '{"entity_id":"{{ entity_id }}","state":"{{ state }}","duration":"{{ duration }}"}'

HA automation.yaml — detect & call n8n

alias: Door Alert → n8n triggers: - trigger: state entity_id: binary_sensor.front_door to: "on" for: "00:05:00" actions: - action: rest_command.n8n_alert data: entity_id: binary_sensor.front_door state: open duration: 5

n8n — Ollama AI Summary + Push to Phone

// Ollama node (or HTTP Request to Gemini) POST http://host.docker.internal:11434/api/chat { "model": "llama3.1", "stream": false, "messages": [ { "role": "system", "content": "Write ONE friendly smart home alert. Max 15 words." }, { "role": "user", "content": "={{ $json.body.entity_id + ' has been ' + $json.body.state + ' for ' + $json.body.duration + ' min.' }}" } ] } // HA node: notify.mobile_app → broadcast intent to Tasker message: command_broadcast_intent data: intent_package_name: net.dinglisch.android.taskerm intent_action: com.example.HA_ALERT_ACK intent_extras: "summary:{{ $('Ollama').item.json.message.content | urlencode }}:String.urlencoded,entity:{{ $json.body.entity_id }}"

Tasker — Intent Received → AutoNotification Buttons

// Profile: Event → System → Intent Received Action Filter: com.example.HA_ALERT_ACK // Task: AutoNotification with action buttons Title: "🏠 Home Alert" Text: %summary Button 1 Label: "Acknowledge" → Task: POST_ACK_WEBHOOK Button 2 Label: "Dismiss" → Task: DISMISS_ALERT // POST_ACK_WEBHOOK task: HTTP Request POST %N8N_URL Body: { "action": "acknowledged", "entity": "%entity" }

✅ 3-Step Verification

1 Manually trigger HA automation via Developer Tools → check n8n log shows Ollama response

2 Phone notification appears within 15 sec with AI-written summary (not a template string)

3 Tap "Acknowledge" → n8n receives follow-up POST → log shows {"action":"acknowledged"}

Part 3

Visual Reference & YouTube Learning Index

📺Watch order: one video per lab session. Build after every watch. Don't binge.

Focus	Channel / Creator	Best For	Search Keywords
Tasker Core	João Dias (joaomgcd) youtube.com/user/joaomgcd	Tasker developer. HTTP Request, variables, plugins, AI Generator, Shizuku, release walkthroughs.	`Tasker HTTP Request event` `Tasker 6.6 features` `Tasker native JSON read`
HA REST	Everything Smart Home (Lewis Barclay) · 247K subs	REST API, automations, DIY integrations. Local-first focus.	`Home Assistant REST API tutorial` `HA webhook trigger 2025`
HA Automations	Home Automation Guy (Alan Byrne) · 113K subs	Practical automations, notify patterns, scripting	`Home Assistant notify android 2025` `HA command_broadcast_intent Tasker`
HA Weekly	Smart Home Junkie (Ed de Tollenaer) · 88K subs youtube.com/c/smarthomejunkie	Weekly HA tutorials. HA Creator Network member.	`Home Assistant webhook local_only` `HA companion app notifications`
HA Official	Home Assistant (official channel)	Release streams, Assist / LLM conversation agent	`Home Assistant Assist AI 2025` `HA conversation process API`
n8n Core	n8n (official, @n8n-io)	Webhook nodes, AI Agent, HA integration, quick-start	`n8n webhook header auth 2025` `n8n AI agent Home Assistant`
n8n Docker	NetworkChuck	"Run n8n for free" — Docker / VPS self-hosting	`NetworkChuck n8n self-hosted`
n8n Deep Dive	CodingEntrepreneurs	"Getting Started with Self Hosting n8n" — 3.5 hr comprehensive	`CodingEntrepreneurs n8n 2025`
AI / Ollama	AI Foundations	AI agents for beginners; Ollama + n8n integration patterns	`Ollama n8n self-hosted 2025` `n8n local LLM stream false`
HA Hardware	The Hook Up · 585K subs digiblurDIY · 74.8K subs	Deep HA + device automation; ESP devices	`Home Assistant AI agent 2025` `HA OpenAI conversation`

🔍 High-Signal YouTube Search Queries

Tasker HTTP Request event inbound server

Home Assistant command_broadcast_intent Tasker

n8n webhook header auth production URL

n8n Ollama Docker host.docker.internal Linux

Home Assistant REST API long lived token

Tasker native JSON read %http_data

n8n Home Assistant node service call

Gemini API x-goog-api-key header Tasker

Part 4

Google Tasks Micro-Plan

⏱️Each task = 15–20 min max. Mark done before moving on. Copy this list to Google Tasks.

📦 Phase 1 — Tasker Foundations (Tier 1)

Open Tasker → Vars tab → find %BATT and %DEVICEID in the built-in list

⏱ 15 min · Goal: know where built-in vars live

Create Task → HTTP Request → GET https://httpbin.org/get → run it → check log for %http_response_code = 200

⏱ 15 min · Goal: first HTTP call confirmed

Change to POST → add header Content-Type:application/json → body {"test":"%BATT"} → POST to https://httpbin.org/post

⏱ 15 min · Goal: see battery value in response

Add Flash action showing %http_data.json.test → confirm it shows the battery number

⏱ 15 min · Goal: native JSON read working

Add If block: if %http_response_code != 200 → Flash "Error: %http_response_code" → test with a bad URL

⏱ 15 min · Goal: error branch fires correctly

🏠 Phase 2 — Home Assistant REST (Tier 2)

HA: Profile → Security → Create Long-Lived Access Token → store in Tasker global %HA_TOKEN

⏱ 10 min · Goal: token created and stored

Tasker: GET http://[HA-IP]:8123/api/ with Authorization:Bearer %HA_TOKEN → confirm {"message":"API running."}

⏱ 15 min · Goal: HA REST connection live

Tasker: GET /api/states/light.YOUR_LIGHT → Flash %http_data.state → shows "on" or "off"

⏱ 15 min · Goal: reading entity state

Tasker: POST /api/services/light/turn_on body {"entity_id":"light.YOUR_LIGHT"} → light turns on

⏱ 15 min · Goal: first real device control from phone

HA: Create webhook automation (local_only: true, ID: tasker-test-001) → POST from Tasker → automation fires

⏱ 20 min · Goal: HA webhook trigger confirmed

HA automation: add notify.mobile_app → command_broadcast_intent → Tasker Intent Received catches it → Flash fires

⏱ 20 min · Goal: HA → phone return path working

⚙️ Phase 3 — n8n Hub Setup (Tier 3)

n8n UI → new workflow → add Webhook node → copy Test URL → click "Listen for test event"

⏱ 15 min · Goal: n8n webhook visible and listening

Tasker POST to Test URL → confirm JSON appears in n8n panel → then read why Production URL is different

⏱ 15 min · Goal: test vs production URL difference locked in

Add Header Auth to webhook → set secret → update Tasker to send X-Webhook-Secret header → still works

⏱ 15 min · Goal: secured webhook

Add "Respond to Webhook" node → body {"ok":true,"reply":"hello from n8n"} → Tasker reads %http_data.reply → Flash shows it

⏱ 20 min · Goal: full synchronous round-trip confirmed

Add built-in HA node → configure credentials (host + token) → call light.turn_on from n8n → light responds

⏱ 20 min · Goal: n8n controlling HA confirmed

Activate workflow → change Tasker URL to Production URL → full end-to-end test → not test mode anymore

⏱ 15 min · Goal: production path live

🤖 Phase 4 — AI Integration (Tier 4)

Store Gemini API key in %GEMINI_KEY → Tasker POST to Gemini generateContent with prompt "Say hello" → read %http_data.text

⏱ 20 min · Goal: direct Gemini call from phone

n8n: add HTTP Request node → POST to http://host.docker.internal:11434/api/chat → stream:false → confirm Ollama response

⏱ 20 min · Goal: Ollama reachable from n8n Docker container

Build Lab 2 step by step (AI Light Commander) — follow Lab 2 blueprint above

⏱ 45 min total, split 2–3 sessions · Goal: voice → light control via AI

🔄 Phase 5 — Return Paths + Capstone (Tier 5)

Enable Tasker HTTP Request event on port 1821 → test with curl from PC → Tasker responds with HTTP Response action

⏱ 20 min · Goal: phone acts as local HTTP server

n8n: after job completes, HTTP Request POST to http://[phone-IP]:1821/ai-result → Tasker receives and Flash result

⏱ 20 min · Goal: async push from n8n to phone over LAN

Build Lab 3 step by step (Smart Alert Processor) — follow Lab 3 blueprint above

⏱ 60 min total, split 3 sessions · Goal: HA → n8n → AI → button notification

🎓 CAPSTONE: single voice command → n8n agent → HA action → AI reply spoken aloud → confirm-or-cancel notification with buttons

⏱ 90 min total · This is the Tier 5 win condition. All 5 tiers complete.

🔑 All secrets live in Tasker globals — never hardcode them: %HA_TOKEN · %N8N_SECRET · %GEMINI_KEY · %HA_HOST · %N8N_URL

Tasker → n8n → HA → AIMaster Training Curriculum

Operational Syllabus & Roadmap

Hands-On Lab Blueprints

🎯 Objective

Triggers, Variables & Actions

Tasker HTTP Request

HA automation.yaml — webhook → broadcast back

Tasker Intent Received Profile

✅ 3-Step Verification

🎯 Objective

Triggers, Variables & Actions

Tasker HTTP Request — JSON Payload

n8n Workflow — Key Node Configs

Tasker — Read Reply & Speak

✅ 3-Step Verification

🎯 Objective

Triggers, Variables & Actions

HA configuration.yaml — rest_command

HA automation.yaml — detect & call n8n

n8n — Ollama AI Summary + Push to Phone

Tasker — Intent Received → AutoNotification Buttons

✅ 3-Step Verification

Visual Reference & YouTube Learning Index

🔍 High-Signal YouTube Search Queries

Google Tasks Micro-Plan

Tasker → n8n → HA → AI
Master Training Curriculum