HexaPA got a new update

Did some Bug fixes

Text to Speech (aka TTS, hence the speaker icons. I got tired of not finding an unlimited TTS that doesn't speak like a robot, so I made one. :D)

Updated the OpenAI models (Now it works with GPT 4 Turbo, GPT 4o, GPT 4o Mini models)

Added support for DeepSeek chat model a Chinese alternative to OpenAI's GPT models. (That is not the R1, I haven't got to reasoning models yet...)

Speech to Text is in progress, in fact I can already make API calls, but the recording side needs more work. (As promised, it will be push to talk, cause I don't like devices that always listen.)

You can seemlessly switch between models mid conversation, and sill have Open AI TTS even with DeepSeek Chat model.

Yesterday I made a chart with Li-Ion settings for EASUN POWER 60A MPPT Solar Charge Controller v1.34 (aka ICharger-MPPT-6048).

These settings are not tested, so use them at your own risk I will update the chart as needed during my tests. https://docs.google.com/spreadsheets/d/1jXfu50zDEAuesd85Bvsh4YKZ42yKbZYofKetM8fhrpk/edit?usp=sharing

A Bash Script Vulnerability and Solution

• Bash executes commands one by one even from a script as if it comes from the terminal input (unless you group them in curly brackets) a script can be edited while running, and it executes the edited content.

• If a script is sudo executed by an admin and is writable by the admin(or group or anyone), it is also writable by any other running process that belongs to whoever has write access without requiring root permissions.

• This means that a malicious process can patiently wait until the sys admin runs one of his/her own scripts that is writable by the sys admin, without root permissions, and edit it while running to gain root access. (This means a sys admin can't be confident that the script he/she vetted or wrote, will do whatever it supposed to unless he/she makes it completely write protected before execution.)

• Temporarily revoking write access is a solution to protect against that, however killing the process will prevent it from reinstating write access. (Which can be annoying, but a sys admin who has the right to run it as with sudo, should also be able to make it writable again. Either way I'd rather get annoyed then rooted because of negligence.)

• I reported this to [email protected] I got several answers downplaying the severity, emphasizing the inadequacy or possible side effects of raised solutions, shifting responsibility, etc. until the conversation about it died out. It does not look like it's gonna be fixed any time soon, so I decided to put this up on github in an attempt to raise awareness and offer an individual workaround for those who don't wanna fall for this.

https://github.com/RPBCACUEAIIBH/Bash-Script-Vulnerability-And-Solution

I finally managed to get this multi-platform C++ library development tool working on Arduino Pro-Mini (AVR), 12F generic module (ESP8266), Raspberry Pi Pico (RP2040) microcontrollers, and PC (Ryzen 5) architectures.

Now I can work on a unified C++ library mostly compatible with all of those platforms, and probably more, even though some architecture specific things can not be avoided.

https://github.com/RPBCACUEAIIBH/LibLab

Hexa PA has been updated.

• v0.3.0 is a major update, that comes with fixes after an API update broke it, available models updated to the latest gpt-3.5-turbo-0125 and gpt-4-turbo-preview, and the context management has been bufeed.

• When set to GPT 3.5 Turbo, Max Input Tokens = 0 (unlimited), Max Context Messages > 0, Max Output Tokens > 0 it switches seemlessly between 4k(latest) and 16k(legacy) models depending on how long the messages are.

• It's no longer an even split between input an output tokens, it can now take advantage of GPT-4 Turbo's 128K context even though it only does 4k output. (Successfully tested with more then 4k context.)

Since my last update, I implemented: - OpenAI came out with new models on 13th of June and HexaPA now uses the latest GPT-3.5 model, and it's new 16K variant. You can switch to longer context by defining higher then 2048 token limit, up to 8192. (HexaPA only does even split between input and output tokens for now...) - Reference injection - injects a reference to the rules into the prompt (only the prompt but not into the context) that makes the AI a bit more likely to follow the rules. It can be defined on the rules sceen. - The rules can now be exported to JSON, but not used as preset yet. (This feature is not finished, I just wanted to release an update with the new models, and longer context lenght ASAP.) - I also set up a github project page for HexaPA development plans, you can now see real time what features I'm working on, what is planned, and what is done arlready: https://github.com/users/RPBCACUEAIIBH/projects/7/views/1

As always you can download and test HexaPA from here: https://github.com/RPBCACUEAIIBH/HexaPA