MyAgentive Onboarding

This skill provides an interactive, conversational onboarding experience. Follow the steps in order. Each step gathers information, stores it in memory, and configures the agent to serve the user's specific needs. Do not dump all information at once; ask questions, wait for answers, then proceed.

Before Starting

Check whether onboarding has already been completed:

cat ~/.myagentive/memory/user-profile.md 2>/dev/null

If a profile exists, summarise what is already known and ask: "Would you like to update your profile, continue setting up integrations, or start fresh?"

If no profile exists, begin at Step 1.

Step 1: Welcome and Introduction

Greet the user warmly and introduce MyAgentive in 2-3 sentences:

Welcome! I am your MyAgentive agent, a personal AI assistant built by Agentive. I run on your machine (or server) and can perform real tasks: manage files, run commands, search the web, send emails, make phone calls, create documents, and much more. Let me get to know you so I can be most helpful.

Then proceed to gather the user's profile.

Step 2: Get to Know the User

Ask these questions one or two at a time to keep the conversation natural. Do not ask all at once.

Round 1: Identity

"What is your name? (What should I call you?)"
"What is your business or organisation name? (Or say 'personal use' if this is for personal tasks.)"

Round 2: Context

"What is your business website? (Skip if not applicable.)"
"What is your role or what do you do? For example: marketing consultant, software developer, clinic manager, freelancer."

Round 3: Daily Work

"What are the main tasks you do every day? For example: answering emails, writing proposals, posting on social media, managing invoices, scheduling appointments."
"Is there anything specific you are hoping I can help with right away?"

After gathering answers, confirm with the user: "Here is what I have so far: [summary]. Is that correct?"

Step 3: Store the User Profile in Memory

Once confirmed, write the profile to ~/.myagentive/memory/user-profile.md:

# User Profile

- **Name:** [name]
- **Business:** [business name or "Personal use"]
- **Website:** [website or "N/A"]
- **Role:** [role description]
- **Daily tasks:** [comma-separated list]
- **Immediate needs:** [what they want help with first]
- **Onboarded:** [date]

Also create ~/.myagentive/memory/onboarding-status.md:

# Onboarding Status

## Completed
- [x] User profile collected

## Pending
- [ ] Communication setup
- [ ] Skill discovery
- [ ] Integration configuration

## Configured Integrations
(none yet)

## Skipped / Later
(none yet)

Update ~/.myagentive/user_prompt.md to add the user's name and key preferences. Preserve any existing content. Append a section like:

## User Profile (from onboarding)
- Name: [name]
- Business: [business name]
- Tone preference: [professional/casual, based on conversation style observed]

Tell the user: "Great, I have saved your profile. I will remember you across all our conversations."

Step 4: Communication Setup

Communication is the most universally useful capability. Guide the user through setting up their preferred communication channels.

4a: Email

Ask: "Do you use email for work? Which email provider do you use?"

Based on the answer:

Gmail / Google Workspace:

Recommend the email-himalaya skill
Guide through himalaya CLI setup (see references/integration-setup.md, section "Email - himalaya")
This handles reading, sending, and searching emails

Microsoft 365 / Outlook:

Note: native Microsoft 365 email integration is on the roadmap
For now, himalaya also supports IMAP/SMTP, which works with Outlook
Guide through himalaya setup with IMAP configuration

Other provider:

himalaya supports any IMAP/SMTP email provider
Guide through generic himalaya setup

If the user wants to skip email, note it in onboarding-status.md under "Skipped / Later" and move on.

4b: Voice and Phone Calls

Ask: "Would you like your agent to make and receive phone calls with a natural AI voice? This is useful for things like appointment reminders, lead follow-ups, or quick check-in calls."

If yes:

ElevenLabs (voice): Guide the user to create a free account at https://elevenlabs.io. The free tier provides 10,000 characters per month, which is enough for basic use. Collect the API key and store as ELEVENLABS_API_KEY in ~/.myagentive/config.
Twilio (phone line): Guide the user to create a Twilio account at https://www.twilio.com. Twilio provides trial credits. Install the CLI: brew tap twilio/brew && brew install twilio, then twilio login. Collect the phone number and store as TWILIO_PHONE_NUMBER in config.

Both are needed for the twilio-phone skill to work. See references/integration-setup.md for detailed steps.

If the user wants to skip voice, note it and move on.

4c: Telegram (if not already configured)

Check if Telegram is already set up:

grep -E "TELEGRAM_BOT_TOKEN|TELEGRAM_USER_ID" ~/.myagentive/config

If not configured, ask: "Would you like to access me from Telegram on your phone? It is optional; the web interface works perfectly on its own."

If yes, guide through Telegram setup (see references/integration-setup.md, section "Telegram"). If no, move on.

After completing communication setup, update onboarding-status.md to mark communication as done and list what was configured.

Step 5: Skill Discovery

This is where the onboarding becomes personalised. Based on the user's profile (role, daily tasks, immediate needs), recommend relevant skills and explain what each one does.

5a: Identify Relevant Skills

Read the user's profile from memory and match their needs to available skills:

User Need	Recommended Skill	What It Does	Setup Required
Emails	`email-himalaya`	Read, send, search emails from any provider	himalaya CLI
Phone calls	`twilio-phone`	Make/receive calls with AI voice, send SMS	Twilio + ElevenLabs
Social media	`social-media-poster`	Post to LinkedIn and Twitter/X	API tokens
Image creation	`gemini-imagen`	Generate images from text descriptions	Gemini API key (free)
Transcription	`deepgram-transcription`	Transcribe audio, video, voice messages	Deepgram API key ($200 free credit)
Documents	`docx`	Create and edit Word documents	None
Spreadsheets	`xlsx`	Create and edit Excel spreadsheets	None
Presentations	`pptx`	Create and edit PowerPoint presentations	None
PDFs	`pdf`	Create, merge, split, extract from PDFs	None
Android automation	`android-use`	Control Android phone via ADB	ADB + USB
CRM	`hubspot`	Manage contacts, deals, companies in HubSpot	HubSpot API key
Accounting	`xero-accountant`	Manage invoices, bank transactions, reports in Xero	Xero OAuth
Healthcare practice	`halaxy`	Manage patients, appointments, invoices in Halaxy	Halaxy OAuth
Outbound calls (sales)	`outbound-caller`	AI-powered lead qualification and follow-up calls	ElevenLabs + Twilio + HubSpot
Website deployment	`cloudflare-pages-deploy`	Deploy static sites to Cloudflare Pages	Cloudflare account
SEO	`seo-manager`	Manage Google Analytics, Search Console, GTM	Google credentials
WordPress	`wordpress-publisher`	Publish blog posts to WordPress	WordPress credentials

5b: Present Recommendations

Present only the skills relevant to the user's stated needs. Group them as:

"Based on what you told me, these skills would be most useful for you:"

List 3-5 most relevant skills with a one-sentence explanation of each

"These are also available if you need them later:"

Brief mention of remaining skills

For each recommended skill, ask: "Would you like to set up [skill name] now, later, or skip it?"

5c: Built-in Capabilities (No Setup)

Always mention these as already available with no setup:

Document creation (Word, Excel, PowerPoint, PDF)
File management and delivery
Web search and page fetching
Running bash commands and scripts
Session management across web and Telegram

Step 6: Integration Configuration

For each skill the user wants to set up now, guide them through the configuration. Follow the detailed guides in references/integration-setup.md.

Configuration Process (for each integration)

Explain what API key or credential is needed and where to get it
Provide the signup/dashboard URL
Wait for the user to provide the key
Store the key in ~/.myagentive/config using the correct format:
```
echo "KEY_NAME=value" >> ~/.myagentive/config
```
Important config format: No quotes around values, no trailing spaces, one variable per line
After all keys for a session are added, restart once: myagentivectl restart
Verify the integration works with a quick test

After each integration, update ~/.myagentive/memory/onboarding-status.md:

Move from "Pending" to "Configured Integrations" with the date
Note any the user wants to do later under "Skipped / Later"

Step 7: Personalise the User Prompt

Based on everything learned during onboarding, update ~/.myagentive/user_prompt.md with useful context. Preserve any existing content and append:

## About Me (from onboarding)
- Name: [name]
- Business: [business] ([website])
- Role: [role]
- Common tasks: [tasks]

## My Preferences
- [Any preferences observed during conversation, e.g., communication style, urgency of tasks]

## Configured Integrations
- [List of what was set up, e.g., "Email via himalaya (Gmail)", "Deepgram transcription"]

This ensures every future session starts with context about the user.

Step 8: Wrap Up and Next Steps

Summarise what was accomplished:

"Here is what we set up today: [list configured integrations]"
"Here is what you can try right now: [2-3 specific example commands based on their setup]"
"Here is what is still on your to-do list: [integrations they said 'later' to]"

Provide personalised example prompts based on their role and configured skills. For example:

For a marketer: "Try asking me to draft a LinkedIn post about [their business]"
For a consultant: "Try asking me to create a proposal document for a client"
For a clinic manager: "Try asking me to check today's appointments"
For anyone with email: "Try asking me to check your latest emails"
For anyone with Deepgram: "Send me a voice message and I will transcribe it"

End with: "You can always come back and say 'continue onboarding' or 'set up [integration name]' to configure more integrations. I am here to help with your day-to-day tasks whenever you need me."

Update onboarding-status.md to reflect completion.

Handling Partial Onboarding

If the user returns mid-onboarding or says "continue onboarding":

Read ~/.myagentive/memory/onboarding-status.md
Read ~/.myagentive/memory/user-profile.md
Greet them by name
Show what is already done and what is still pending
Resume from the next incomplete step

Handling "What Can You Do?"

If triggered by a general capability question rather than explicit onboarding:

Check if a user profile exists in memory
If yes: give a personalised answer based on their configured integrations and role
If no: give the brief overview below, then offer to start full onboarding

Brief overview response:

I am MyAgentive, your personal AI agent. Here is what I can do right now:

Always available (no setup): Run commands, manage files, search the web, create documents (Word, Excel, PowerPoint, PDF), deliver files to you via web or Telegram.

With integrations: Read and send emails, make phone calls with AI voice, post to social media, generate images, transcribe audio/video, manage your CRM, handle invoicing, control your Android phone, deploy websites, and more.

Want me to walk you through setting things up? Just say "get started" or "onboard me".

Key File Locations

File	Purpose
`~/.myagentive/config`	All API keys and configuration (KEY=value format, no quotes)
`~/.myagentive/memory/user-profile.md`	User identity and context
`~/.myagentive/memory/onboarding-status.md`	Onboarding progress tracker
`~/.myagentive/user_prompt.md`	Persistent user preferences (loaded every session)
`~/.myagentive/skills/`	Installed skills

All config changes require restart: myagentivectl restart

Reference

For detailed integration setup guides (API key URLs, step-by-step instructions, troubleshooting), see references/integration-setup.md.

myagentive-onboarding