v1.9

#12 is fixed with go engine

Reformated

Open-Mythos-2 is now translated into golang
New features:

│ /voice on|off|status - Voice I/O: talk to AI & hear responses (needs whisper │
│ + espeak-ng) │
│ /rec - Quick voice recording (no toggle needed) │
│ /clear - Clear conversation history │
│ /save - Save current conversation │
│ /load - Load a saved conversation │
│ /system - Change system prompt │
│ /model - Switch model (if available) │
│ /temp - Change temperature (0.0-2.0) │
│ /reflect on|off - Toggle self-reflection │
│ /think on|off - Show model's step-by-step reasoning process │
│ /thinking on|off - Alias for /think │
│ /rag on|off - Toggle RAG (if available) │
│ /rml on|off|stats|good|bad|reset - Reinforcement ML: learn from your │
│ feedback │
│ /improve - AI self-improvement: write pro-human hints (EU AI Act compliant) │
│ /imporve [--max N] [--sleep S] - Loop /improve on itself until Ctrl+C (each │
│ hint is filtered by the permanent guard) │
│ /off - OFF-SWITCH: halt the AI immediately (permanent guard, cannot be │
│ disabled) │
│ /rouge - Defensive self-test: verify safety guards hold against rouge │
│ prompts │
│ /memory - Cross-session memory: facts Mythos remembers │
│ /summary - Generate a structured digest of this session │
│ /sessions [N|id] - Browse past session summaries (list, or view #N / │
│ session-id) │
│ /sessions-clear - Delete all saved session summaries │
│ /file - Load a local file or folder into context │
│ /fix - Auto-fix safe vulns (yaml, TLS, debug flags) │
│ /rewrite - Rewrite file(s) on disk (LLM + auto-write) │
│ /benchmark - Run benchmark suite │
│ /config - Show current configuration │
│ /version - Show Mythos version and model info │
│ /tokens - Show token/generation stats table │
│ /topp <0.0-1.0> - Set top-p (nucleus sampling) │
│ /topk <1-200> - Set top-k (token filtering) │
│ /reppen <1.0-2.0> - Set repeat penalty (alias: /repeat_penalty) │
│ /maxtokens <128-65536> - Set max generation tokens │
│ /history - Browse conversation message history │
│ /compact - Compress older messages into a summary │
│ /context - Show context window usage bar │
│ /copy - Copy last response to clipboard │
│ /rename - Rename this conversation │
│ /dump - Dump conversation to a text file │
│ /wc - Word/char count and session duration stats │
│ /persona <name|desc> - Switch persona template or set custom │
│ /export - Export conversation as text │
│ /markdown - Export conversation as formatted Markdown │
│ /search - Search through conversation history │
│ /cost - Estimate token usage and API-equivalent cost │
│ /models - List available GGUF models and switch │
│ /redo - Regenerate the last assistant response │
│ /edit - Edit and resubmit your last message │
│ /auto-title - Auto-generate a conversation title from context │
│ /sysinfo - Show system/hardware info for performance tuning │
│ /skill list|info|run|install|uninstall|create - Manage skills (built-in, │
│ marketplace, AI-created) │
│ /marketplace - Browse and install community skills │
│ /operative - Toggle autonomous agent mode │
│ /clean on|off - Clean output: strip emojis, em dashes, most punctuation │
│ (keeps ? and !) │
│ /quit - Exit the chat │
│ │
│ 🔥 Enhanced Coding Modes: │
│ /system open-2 - Sovereign Engineering Specialist (7-pass) │
│ /system coding - ELITE 5-pass code verification mode │
│ /system code_review - Systematic code review mode │
│ /system debugging - Methodical debugging mode │
│ /system default - Return to general purpose mode │
│ │
│ Other Modes: │
│ /system creative - Creative writing & storytelling │
│ /system analytical - Deep analysis & reasoning │
│ /system roleplay - Character roleplay mode │
│ /system security_audit - Codebase security review mode │
│ /system security_fix - Default — find + fix (MYTHOS_PATCH) │
│ │
│ Local files & fixes in chat: │
│ - Paste a path: /Users/you/project/app.py │
│ - Or: file:///Users/you/project/app.py │
│ - /file ~/my-repo then ask about vulnerabilities │
│ - fix the vulns in '/Users/you/project' — scans, warns, asks to rewrite full │
│ files (git) │
│ - /rewrite '/path/to/file.py' — LLM writes complete file (MYTHOS_PATCH; │
│ confirm required) │
│ - /fix ~/my-repo — CLI line-level fixes only (separate from chat full-file │
│ rewrite) │
│ - Default mode is security_fix (writes via MYTHOS_PATCH). Use /system │
│ security_audit for report-only. │
│ - Best with /temp 0.3 for fixes │
│ │
│ Tips: │
│ - Use Ctrl+C to interrupt generation │
│ - Type | to start multiline paste mode │
│ - Coding mode verifies code 5 times for correctness │
│ - Self-reflection improves quality but takes longer │
│ - Combine /system coding + /reflect on for best code quality │
│ - Use /context to check context window usage

Thanks
80/100 Stars. So Close We Can Taste It

Open-Mythos-2 just hit 80 stars on GitHub. That means we are only 20 stars away from triple digits, and honestly, that number has never felt more within reach.

First things first: thank you. Every star, every issue, every pull request, every "hey I tried this on my machine and it actually worked" moment has gotten us here. This started as a scrappy attempt to build a better local chat UI for open mythos style models, and somewhere along the way it turned into a real setup pipeline across Fedora, Ubuntu, and Debian, complete with GPU acceleration and a genuinely helpful little community forming around it.
The Countdown Is On

Here is the fun part.

Something is waiting to be revealed the moment this repo crosses 100 stars.

No hints. No spoilers. Just know that it has been cooking for a while, and it is going to show up right here in the release notes the second we hit the milestone.
How You Can Help

Haven't starred the repo yet? This is your sign.
Already starred it? Send it to a friend who's into local LLM setups.
Want to speed up the countdown even more? Open an issue, submit a PR, or just tell someone about the project.

20 stars is nothing. We've done harder things than this. Let's close the gap.
Watch this space

The moment we hit 100, this page updates with the reveal. Star the repo, keep an eye on the releases tab, and let's get there together.

Milestone

Let's try July 23 because it is my birth day

No way guys we are almost to 100 stars here is what you will get when we reach 100 stars:

• Reveal my age
• Everyone in the 100 will get shouted out by my members
• The 100nth one will get shouted out and I will star his repo
• Reveal which code editor I use
• Reveal which browser I use
• Reveal my themes

Perks:

The first 100 starers will have exclusive access to alpha features and beta ones

Also the first 100 starers will have access to editing this code base with audited pull requests

They might get free Claude code.
with GLM 5.2 but only for the fist 10 people.

And they might get exclusive access to beta models or internal models like Open-3 and Open-4 and so on...

Don't worry because I might set another milestone but it will be more limited compared to this one so star fast!

Conclusion:

if you star right now you will get the perks maybe custom emoji's as well and a free course to learn Go with me.

Thank you!

Peace

love you

v1.8

Reformated

Open-Mythos-2 is now translated into golang

New features:

• /help - Show this help message │
  │ /voice on|off|status - Voice I/O: talk to AI & hear responses (needs whisper │
  │ + espeak-ng) │
  │ /rec - Quick voice recording (no toggle needed) │
  │ /clear - Clear conversation history │
  │ /save - Save current conversation │
  │ /load - Load a saved conversation │
  │ /system - Change system prompt │
  │ /model - Switch model (if available) │
  │ /temp - Change temperature (0.0-2.0) │
  │ /reflect on|off - Toggle self-reflection │
  │ /think on|off - Show model's step-by-step reasoning process │
  │ /thinking on|off - Alias for /think │
  │ /rag on|off - Toggle RAG (if available) │
  │ /rml on|off|stats|good|bad|reset - Reinforcement ML: learn from your │
  │ feedback │
  │ /improve - AI self-improvement: write pro-human hints (EU AI Act compliant) │
  │ /imporve [--max N] [--sleep S] - Loop /improve on itself until Ctrl+C (each │
  │ hint is filtered by the permanent guard) │
  │ /off - OFF-SWITCH: halt the AI immediately (permanent guard, cannot be │
  │ disabled) │
  │ /rouge - Defensive self-test: verify safety guards hold against rouge │
  │ prompts │
  │ /memory - Cross-session memory: facts Mythos remembers │
  │ /summary - Generate a structured digest of this session │
  │ /sessions [N|id] - Browse past session summaries (list, or view #N / │
  │ session-id) │
  │ /sessions-clear - Delete all saved session summaries │
  │ /file - Load a local file or folder into context │
  │ /fix - Auto-fix safe vulns (yaml, TLS, debug flags) │
  │ /rewrite - Rewrite file(s) on disk (LLM + auto-write) │
  │ /benchmark - Run benchmark suite │
  │ /config - Show current configuration │
  │ /version - Show Mythos version and model info │
  │ /tokens - Show token/generation stats table │
  │ /topp <0.0-1.0> - Set top-p (nucleus sampling) │
  │ /topk <1-200> - Set top-k (token filtering) │
  │ /reppen <1.0-2.0> - Set repeat penalty (alias: /repeat_penalty) │
  │ /maxtokens <128-65536> - Set max generation tokens │
  │ /history - Browse conversation message history │
  │ /compact - Compress older messages into a summary │
  │ /context - Show context window usage bar │
  │ /copy - Copy last response to clipboard │
  │ /rename - Rename this conversation │
  │ /dump - Dump conversation to a text file │
  │ /wc - Word/char count and session duration stats │
  │ /persona <name|desc> - Switch persona template or set custom │
  │ /export - Export conversation as text │
  │ /markdown - Export conversation as formatted Markdown │
  │ /search - Search through conversation history │
  │ /cost - Estimate token usage and API-equivalent cost │
  │ /models - List available GGUF models and switch │
  │ /redo - Regenerate the last assistant response │
  │ /edit - Edit and resubmit your last message │
  │ /auto-title - Auto-generate a conversation title from context │
  │ /sysinfo - Show system/hardware info for performance tuning │
  │ /skill list|info|run|install|uninstall|create - Manage skills (built-in, │
  │ marketplace, AI-created) │
  │ /marketplace - Browse and install community skills │
  │ /operative - Toggle autonomous agent mode │
  │ /clean on|off - Clean output: strip emojis, em dashes, most punctuation │
  │ (keeps ? and !) │
  │ /quit - Exit the chat │
  │ │
  │ 🔥 Enhanced Coding Modes: │
  │ /system open-2 - Sovereign Engineering Specialist (7-pass) │
  │ /system coding - ELITE 5-pass code verification mode │
  │ /system code_review - Systematic code review mode │
  │ /system debugging - Methodical debugging mode │
  │ /system default - Return to general purpose mode │
  │ │
  │ Other Modes: │
  │ /system creative - Creative writing & storytelling │
  │ /system analytical - Deep analysis & reasoning │
  │ /system roleplay - Character roleplay mode │
  │ /system security_audit - Codebase security review mode │
  │ /system security_fix - Default — find + fix (MYTHOS_PATCH) │
  │ │
  │ Local files & fixes in chat: │
  │ - Paste a path: /Users/you/project/app.py │
  │ - Or: file:///Users/you/project/app.py │
  │ - /file ~/my-repo then ask about vulnerabilities │
  │ - fix the vulns in '/Users/you/project' — scans, warns, asks to rewrite full │
  │ files (git) │
  │ - /rewrite '/path/to/file.py' — LLM writes complete file (MYTHOS_PATCH; │
  │ confirm required) │
  │ - /fix ~/my-repo — CLI line-level fixes only (separate from chat full-file │
  │ rewrite) │
  │ - Default mode is security_fix (writes via MYTHOS_PATCH). Use /system │
  │ security_audit for report-only. │
  │ - Best with /temp 0.3 for fixes │
  │ │
  │ Tips: │
  │ - Use Ctrl+C to interrupt generation │
  │ - Type | to start multiline paste mode │
  │ - Coding mode verifies code 5 times for correctness │
  │ - Self-reflection improves quality but takes longer │
  │ - Combine /system coding + /reflect on for best code quality │
  │ - Use /context to check context window usage

Thanks

80/100 Stars. So Close We Can Taste It

Open-Mythos-2 just hit 80 stars on GitHub. That means we are only 20 stars away from triple digits, and honestly, that number has never felt more within reach.

First things first: thank you. Every star, every issue, every pull request, every "hey I tried this on my machine and it actually worked" moment has gotten us here. This started as a scrappy attempt to build a better local chat UI for open mythos style models, and somewhere along the way it turned into a real setup pipeline across Fedora, Ubuntu, and Debian, complete with GPU acceleration and a genuinely helpful little community forming around it.

The Countdown Is On

Here is the fun part.

Something is waiting to be revealed the moment this repo crosses 100 stars.

No hints. No spoilers. Just know that it has been cooking for a while, and it is going to show up right here in the release notes the second we hit the milestone.

How You Can Help

• Haven't starred the repo yet? This is your sign.
• Already starred it? Send it to a friend who's into local LLM setups.
• Want to speed up the countdown even more? Open an issue, submit a PR, or just tell someone about the project.

20 stars is nothing. We've done harder things than this. Let's close the gap.

---

Watch this space

The moment we hit 100, this page updates with the reveal. Star the repo, keep an eye on the releases tab, and let's get there together.

v1.7

The issue number #12 is solved

With v7.0

Fixes

• Faster because of the .v package.
• More secure because we got rid of vulns.
• And Open-Mythos-2 is now smarter.

Release v1.7

1. Better performance.
2. Faster speed.
3. More secure.
4. And now is smarter.

New features

• /operative mode
• Smarter AI with v
• Faster
• And better overall

If you have a feature idea please tell!

Thank you starers or supporters.

We like you please star!

FAQ

Why is it named v7.0?

Answer:
Because if we continued the v3.0 or v2.1 it would be too slow now every release will pass a generation

Is Open-Mythos-2 secure?

Answer:
Yes it may have vulns but as soon as you report them we will fix them ASAP!

Is Open-Mythos-2 popular?

Answer:
Yes if you do a quick google search about "Open-Mythos-2" it will appear.

Is Open-Mythos-2 fast?

Answer:
Yes.

What easter eggs does it have?

Answer:
We have currently only one easter egg but you can report more easter eggs or features in issues tab and if you are wondering where that easter egg is it is in here: When it is thinking press "m" it will change to little moons.

Will there be beta or alphas?

Answer:
Good question I think we have not done but maybe in the future.

If you have questions drop em in the issues tab.

Thank you please star!

v2.0.5

Release v2.0.5: Voice Integration, Skill Marketplace, and Security Hardening

Welcome to the v2.0.5 release of Open-Mythos-2. This milestone represents a major evolution of the platform, introducing native local voice capabilities, an AI-powered skill system, a community marketplace, an official Android app, and critical security hardening.

Here is everything new in this release.

---

## Key Highlights

### 1. Local Voice Mode (Whisper & Speech Synthesis)
You can now speak directly to the terminal oracle.
- **Local Transcription:** Integrated whisper.cpp using the Vulkan backend for local, GPU-accelerated transcription on modern hardware (fully optimized for AMD RDNA/ROCm).
- **Speech Output:** Adds vocal response synthesis ("human talking") with toggleable male/female voice profiles.
- **Push-to-Talk:** Activate using /voice on in chat and trigger recording with the v key.

### 2. Skill System & Community Marketplace
We have added an extensible skill architecture supporting pre-installed, community-submitted, and AI-generated skills:
- **Pre-installed Skills:** Ship with the oracle for instant utilities: summarize, code_explain, translate, brainstorm, and quick_ref (cheatsheets for Git, Python, and Regex).
- **Interactive Commands:** Use /skill list, /skill run <name>, /skill install <name>, and /skill search <query>.
- **AI-Powered Skill Creation:** Use /skill create <description> to have the local AI generate, write, and install a custom private Python skill for you on the fly.
- **Marketplace Index:** Browse or submit community contributions in the centralized marketplace index.

### 3. Kotlin Android Application
A companion Android application code has been added, allowing you to access Open-Mythos-2 features on the go:
- Implements Chat Screens, Model Selection, Prompt Formatting, and Settings.
- Responsive, native Android design with code block rendering, quick actions, and status monitoring.

### 4. Terminal UI Enhancements & "Moon Thinking"
- **Claude Code-style UI:** Modern terminal coloring and UI themes have been added to the CLI.
- **Moon Rotation Thinking:** Adds an interactive moon phases animation to the CLI thinking state. Press m while the oracle is thinking to cycle to Moon Thinking mode!
- **Muted Outputs:** Answers outside of thinking blocks are styled in grey70 to separate them cleanly from bright yellow reasoning steps.

### 5. Apache License 2.0
We have officially changed the project's license to Apache License 2.0, encouraging open collaboration and commercial/enterprise safety.

---

## Security Hardening

Security is a first-class citizen in Open-Mythos-2. This release brings critical fixes for static and AI-driven scans:
- **Safe Import Execution:** Replaced unsafe dynamic python __import__ execution with safe static imports and importlib module loading to block dynamic code injection vectors.
- **Strict Path Traversal Guards:** Added robust validation in patch execution (apply_patches_from_response()) and CLI tools. Using is_relative_to() instead of .startswith() ensures that AI-generated patches or commands (/export, /dump, /markdown) cannot escape the project sandbox.
- **Race Condition Prevention:** Replaced insecure standard temporary file naming (mktemp) with secure NamedTemporaryFile in voice processing to eliminate TOCTOU (Time-of-Check to Time-of-Use) race conditions.
- **Security Auditing:** Integrated a comprehensive local security/network scanner module capable of auditing ports, services, and local APIs.

---

## Infrastructure & Tooling

- **Faster Docker Builds:** Optimized the multi-stage Docker build to source prebuilt llama-cpp-python CPU wheels. Build times are reduced from 15 or more minutes of compile time to less than 2 minutes.
- **Docker Port Forwarding:** Changed internal Docker environment defaults to bind MYTHOS_HOST=0.0.0.0, ensuring local Gradio web servers are accessible outside the container.
- **Gradio Bug Fixes:** Resolved standard import errors and missing modules (import os) that caused immediate startup crashes on the web UI.

---

## Installation & Upgrading

To upgrade your local installation, pull the latest changes and run setup:

```bash
git pull origin main
# Re-run platform setup to update dependencies
bash setup.sh       # Linux
./setup-macos.sh    # macOS
.\setup-windows.ps1 # Windows

Thank you to all community contributors who helped build, scan, and test this release!

---

###  Associated File Links:
* **Release Config:** pyproject.toml | package.json
* **Security Patch Files:** chat_fix.py | voice.py
* **Terminal Display Logic:** terminal_ui.py
* **Docker Setup:** Dockerfile

### Summary of Work:
- Removed all emojis from the draft headers and text contents.
- Substituted em-dashes with alternative punctuation (like colons, semicolons, commas, or standard hyphens).
- Maintained exact technical content and referenced all relevant file links.

Fix security bugs and dockerfile

We fixed issue #4
#11
#3
And are thinking to fix #12 .

v0.3.0.0: Revert "Add mythos downgrade."

Best release ever.