Files
Lee Hanken 7c8efd2228 Initial commit: HPR Knowledge Base MCP Server
- MCP server with stdio transport for local use
- Search episodes, transcripts, hosts, and series
- 4,511 episodes with metadata and transcripts
- Data loader with in-memory JSON storage

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-10-26 10:54:13 +00:00

214 lines
14 KiB
Plaintext

Episode: 4239
Title: HPR4239: Android Tasker and Automation
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr4239/hpr4239.mp3
Transcribed: 2025-10-25 21:53:50
---
This is hacker public radio episode 4239 for Thursday the 31st of October 2024.
Today's show is entitled, Android Tasker and Automation.
It is hosted by Operator and is about 15 minutes long.
It carries an explicit flag.
The summaries Operator talks about Android Automation and Tasker.
You are listening to a show from the Reserve Q.
We are airing it now because we had free slots that were not filled.
This is a community project that needs listeners to contribute shows in order to survive.
Please consider recording a show for hacker public radio.
Hello everyone, welcome to another episode of hacker public radio with your own operator.
So today should be a pretty quick one.
I am going to talk about Tasker which I probably have talked about before.
Current usage and limitations and experience with it before.
Now Tasker is an Android specific app that allows you to do all kinds of crazy stuff, especially
if you have Admin or root on your phone.
So it goes in order of if you have root, you can do a lot of things.
You basically can do whatever you want.
If you don't have root and you have ADB which is ADB debugging bridge, you can turn that
on and have a little bit more useful functionality, but it's significantly reduced because there's
only so much you can do with ADB, the debug bridge, which my workflow does not have that.
It doesn't allow the ADB debugging, nor do they allow for party apps.
So I did an episode on this, I think already, but you can use user lane, user lane will
let you spin up a Linux distro and do whatever you want.
It's quite buggy and you can't really do kernel stuff whether it uses the ARM kernel.
But that's pretty much how you can do like you do downloaders or whatever you need to run
command line stuff or if you need an app, you can run whatever you want to run in that little
container.
So that's how we've gotten around that.
Now what I have now is I wanted to find a way to make a AI assistant work, an AI voice
prompt to the assistant.
So I can say like a Google and it will bring up a prompt and that prompt will go to chat
GPT or it will go to a large language model locally or a mile and fine tune model.
That's in the works.
And the fine tuning is not quite idiot proof and I don't have the GPU to do that and waiting
for that to decide on that.
If I should do that, I also want to do voice cloning, but anyways I'd address the in-state
is to have a personal assistant on my wife's phone and my phone that will help me with
time management, communication that will relay messages for me, it will interpret tone
with communications because I have problems communicating.
So the idea is that to make basically an automation assistant that will help me communicate
and manage my time.
So it will say, oh look, looks like you double booked your personal calendar with your
work calendar.
You can then sanitize, of course, and separate, but at least being able to say, oh, you got
a busy status here and your personal calendar does, you know, is conflicting with this
calendar or your wife's calendar or another calendar, I've got four or five calendars
across my setup.
So that's what I want.
And I try tasker, tasker now lets you do things like hijack the Google voice prompt
essentially.
This is without admin, it is very clunky tasker is not a app for the faint of heart, it
is for the hacker, it is for the, you know, a pretty advanced user now.
With that said, if you're doing a simple task, like, for example, the reason I'm doing
this one is I noticed that I'd have to turn on, do not disturb and I have it for a max
setting of an hour or two hours and that way I'm not setting an on, do not disturb and
forgetting that it's on, do not disturb and then missing a bunch of messages from somebody
that I need to respond to.
I use that along with repeating notifications, which is basically an app that if you don't,
the beeps, you get a notification every two minutes it keeps repeating for whatever
I'd be specified.
So I have teams and outlook every two minutes, ping me if I get a notification.
So if I'm not paying attention or if I'm gambling or rambling, rambling, it will repeat
that notification and add nauseam until I view it.
Now with that said, when I'm at the office, I'm at the office, I don't need to be, you
know, I'm not doing other things, I'm multitasking generally, I'll be rambling and I do need
alarms for my meetings, but I set my alarms for meetings, I do not use notifications
because that's not enough for me, even an audio audible notification is not enough, I'll
snooze it a million times, I'm actually supposed to be walking right now.
So with that said, between the alarms calendar, what I wanted was an easy way to put myself
on, do not disturb without having to hit it every hour.
So I was hit an hour and then it would start going off again, and then I'd be working
and I'd hit it another hour, this started going off again.
And I'd have like six SMS messages, I'd have like two or three SMS messages and I'd
have like a bunch of discord messages and maybe I'd have a couple of team messages
that I knew weren't grateful, they were, you know, from something else.
And maybe even something in my mailbox, which is heavily filtered for work.
But there are all two cares, my own alarms are the only actionable things that I have,
and I have like five different six, seven, eight different alarms for personal tasks and
work tasks and whatever, and I use that to manage everything because you can override
do not disturb with alarm, so that everything will be quiet except for the do not disturb
or the alarms.
So what I wanted was an easy way to manage that.
And I remembered at some point in time, there was a flip, they call it flip to, or flip
to SSHH, essentially it's a very simple query and I'll say it, I don't even need to post
it in the show notes or anything.
It's quite simple.
It's basically a beeps and waits and it says do not disturb mode alarms.
I put it on mode alarms because it was everything and I was putting do not disturb for the alarms
too, which I do not want my alarms to go off.
So I put do not disturb alarms and it functions on the orientation face down and it looks
like it waits, you know, 300 milliseconds.
And then when it's not doing that on the exit, it says do not disturb, allow, all, wait
and then beep again.
So when I put my phone down, it'll go on do not disturb face down, it'll go on do not
disturb and I don't have to worry about managing or repeating or whatever, and I know mentally
if I'm at home, I don't want to put my phone face down because I might, you know, it might
be an hour before I, you know, do something or an hour and 30 minutes, maybe I go out
to lunch and I come back or whatever.
So I could use it, I don't know, we'll see how it goes.
But anyways, Tasker can also, like I said, hijack your Google voice prompts your Google
Assistant.
What I realized is that the Google app for my work phone again won't let me update it
for whatever reason recently.
I don't know why that is, but I probably privacy or something.
And now I can't use Google Assistant.
But the idea is that it would hijack that and you could say, okay, Google, orange, tell
me the weather in Wisconsin and I'm, you know, whatever, and it will, it will, it will
use JetGPT to do that.
Of course, JetGPT doesn't know the weather in Wisconsin right now.
That's where I was going to rewrite the script to work for Akaji Web Search API thing or
make my own API, which is Internet Search plus AI, which is phenomenal.
Essentially, what do you can do with JetGPT right now?
But you have to pay $3 a month and they, you know, not very focused on privacy.
So I think all of these actually end up going through just opening the way I have them
set up.
But for me, I wanted the Internet Search plus the current plus AI model.
So it doesn't do any good to have an AI model unless you have the Internet.
So that's why I wanted to replace Google Assistant with Akaji, Google Akaji Assistant Beta
or whatever they call it.
It's all in beta.
So that's where I'm at today.
I think that pretty much covers most of the stuff I've done in Tasker.
Back when it first came out, probably 10 years ago, it was the same sort of thing I wanted
to automate Google Voice stuff in Google Voice is not necessarily an SMS client.
It's also not really a phone.
I think it registers as a phone app, but it does not register as a SMS app.
So you can't attach it to the traditional things you can attach SMS messages to or SMS
clients to.
What I wanted was to be able to say, hey, Google, you know, send a text to Kathy.
I need to remember to mow the line, whatever.
And it would create that SMS messages and send it out or it would call my wife or whatever.
I think calling works.
But anyways, that was what I was using it for.
I never could figure out.
I never could figure out the essential local API, what you're doing, like essentially local
like Java function calls to like call classes and stuff.
You're basically sort of doing coding, Java coding, sort of.
You're calling these methods and classes to like trigger stuff.
And I can never really figure it out.
The Google assistant hacking, hijacking thing works the same way.
You run the setup and you say, okay, Google.
And it's like, oh, I've detected the function call.
It's okay, Google.
It's, you know, Google assistant underscore 8348297 and whatever that is, is some UID
or some ID that, you know, that gets passed as that user, whoever, for whatever reason,
is a different.
Excuse me, a unique kind of thing from user to user.
That's what I used before.
I constantly, I just saw a reinstall tasker.
It's, I want to say like $15 or something.
I don't know how much it is, five bucks, a $18, it's not, I don't think remember it being
very expensive.
Now you get into auto apps and I think auto notification, those have a cost around them,
something's beeping.
And my phone is not on, do not disturb.
So I'm getting stuff, I got a 230 here in 15 minutes.
10 minutes.
So that's all I can say about tasker, I've hopped and rambled for about 10 minutes.
What I'm looking for is a search, a voice assistant on Android.
So I can say, okay, Google, and it will do internet search and AI.
So I don't know if chat GPT has something, I would be willing to pay another $300 for
that.
One time price of $300.
It was the same price as opening AI, which you get way more models.
You get all the, with Kaji, you get all the clawed stuff and all the other one and the
opening AI stuff.
So you get access to far more, plus, you know, it's more privacy based in the budget
smart dudes.
So anyways, it's pretty much where I'm at.
And I'm looking for a voice assistant that will let me do whatever my future state is
probably going to be a tasker app that calls a function to talk to local language model
and might say local, I mean, probably a server on the internet that connects to my house
that will use local language model of myself and possibly, you know, a fine-tune model based
on my own data, such as SMS messages, email messages, parsing that out between the
difference between myself and the reply to is going to be difficult, but it's feasible
that I could pull all that in and pull all my SMS messages in, which would be I think
pretty easy over Google Voice.
I mean, I've got probably 20 years, 10 years of SMS stuff in Google Voice.
Pull that all into a fine-tune model and then use that to define a hearty factor.
So the idea would be, you know, zero would be completely unfiltered, you know, I would
never use it, you know, if I'm zero to 10, five would be like, I'm talking to my wife.
So it would do grammar checks, spell check, and it would be a little bit cleaned up.
And then, for example, maybe like 10 would be your nine would be like more leaning towards
the language model and it would have so much of my personality in it.
You can do personality types and whatever they call them, Myers-Briggs personality types
and whatever the other name is, you can do with easily with language models.
So you can say, act in a Myers-Briggs personality of ENFJ in a tri-type integram of 729.
So my Myers-Briggs in tri-type integram is E as an echo in as an Nancy, I don't know
what the phonex is, the EN, F as in Foxtrot, J as in middle of my bro, J is, and then 729
and you can say that will kind of be the closest you can get to, like making a personality
of yourself for like a generic, large language model.
But you could do that along with the line tuning and then I would have it taught to my calendar
and sync all that and taught to my external calendar and sync all that, manage all that,
and then manage communications, not work because it's sensitive, but manage my personal communications
and reword locally, maybe my, you know, work communications, maybe something running on my laptop
or something, I don't know.
But the idea would be to help with that communication and then time management.
But anyways, speaking of time management, we're at 14 minutes, I've got to be on a call here
and I need to do some stair climbing to get some exercise in.
And if anybody knows anything about that, we'll kind of go from there.
You have been listening to Hacker Public Radio at HackerPublicRadio.org.
Today's show was contributed by a HBR listener like yourself.
If you ever thought of recording podcasts, click on our contribute link to find out how easy it leads.
Hosting for HBR has been kindly provided by an onsthost.com, the internet archive and our sync.net.
On this otherwise status, today's show is released on our Creative Commons
Attribution 4.0 International License.