⚡ p-e-w/heretic is making waves. Here's the full picture.

🔗 https://github.com/p-e-w/heretic
📝 Fully automatic censorship removal for language models
──────────────────────────────

The Heretic GitHub repository is a cutting-edge project that enables fully automatic censorship removal for language models. This innovative tool combines advanced directional ablation, also known as "abliteration," with a TPE-based parameter optimizer powered by Optuna. The result is a decensored model that retains as much of the original model's intelligence as possible, without requiring any human effort or understanding of transformer internals.

Key Features:
- Automatic censorship removal
- Supports most dense models, including multimodal and MoE architectures
- Compatible with model quantization using bitsandbytes
- Provides features for research into model internals and interpretability

Usage:
To get started, simply install the heretic-llm package and run the heretic command with the desired model. For example:

pip install -U heretic-llm
heretic Qwen/Qwen3-4B-Instruct-2507

Technical Highlights:
- Employs PaCMAP projection for residual vector analysis
- Offers options for generating plots and printing residual geometry details
- Supports a variety of configuration parameters for customization

Audience:
Heretic is designed for anyone interested in working with language models, from researchers to developers and enthusiasts. The tool's automatic nature makes it accessible to those without extensive knowledge of transformer internals.

In summary, Heretic is a powerful tool that makes it easy to remove censorship from language models, and with its cutting-edge features and ease of use, it's a game-changer for the field - decensor your models with just a few clicks!

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

📌 Spotted on GitHub Trending: EveryInc/compound-engineering-plugin — let's break it down.

🔗 https://github.com/EveryInc/compound-engineering-plugin
📝 Official Compound Engineering plugin for Claude Code, Codex, Cursor, and more
──────────────────────────────

The Compound Engineering Plugin is designed to make each unit of engineering work easier than the last, inverting the traditional development process that often accumulates technical debt. This plugin provides AI skills and agents to streamline the development workflow, focusing on planning, review, and knowledge codification to ensure high-quality changes.

The core features include:
- /ce-brainstorm and /ce-plan for thorough planning
- /ce-code-review and /ce-doc-review for catching issues
- /ce-compound for codifying knowledge

To get started, run /ce-setup in any project, and explore the 37 skills and 51 agents available. The plugin can be installed via various platforms, including Claude Code, Cursor, Codex, GitHub Copilot, Factory Droid, Qwen Code, OpenCode, Pi, Gemini, and Kiro, using specific installation commands.

In summary, the Compound Engineering Plugin is a game-changer for developers, making each change easier than the last by compounding knowledge and streamlining the development process. Install it now and start coding smarter, not harder!

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

🎰 Welcome Bonus 1200% — Maczo Crypto Casino
🎮 Crypto exchange · Sports · Live casino — all in one place
💳 USDT instant deposit & withdrawal
→ https://tglink.io/c07f00ebd9d268

⚡ TauricResearch/TradingAgents is making waves. Here's the full picture.

🔗 https://github.com/TauricResearch/TradingAgents
📝 TradingAgents: Multi-Agents LLM Financial Trading Framework
──────────────────────────────

The TradingAgents framework is a cutting-edge, open-source tool for simulating real-world trading firms using multi-agent systems and large language models (LLMs). Its primary purpose is to facilitate research into the dynamics of financial markets and the development of more sophisticated trading strategies.

Key features of TradingAgents include support for multiple LLM providers (e.g., OpenAI, Google, Anthropic), modular architecture for easy customization, and a user-friendly CLI for interacting with the framework.

To get started with TradingAgents, users can clone the repository, install the package, and launch the interactive CLI. The framework also provides a Python API for more advanced usage.

TradingAgents is designed for researchers and developers interested in exploring the applications of LLMs in financial markets.

Overall, TradingAgents offers a powerful tool for simulating and analyzing complex trading scenarios, making it an exciting development for anyone interested in the intersection of finance and AI: TradingAgents is revolutionizing the way we approach financial modeling and trading strategy development.

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

📌 Spotted on GitHub Trending: revfactory/harness — let's break it down.

🔗 https://github.com/revfactory/harness
📝 A meta-skill that designs domain-specific agent teams, defines specialized agents, and generates the skills they use.
──────────────────────────────

Harness is a team-architecture factory for Claude Code, enabling the creation of complex agent teams and skills from simple domain descriptions. With six pre-defined team-architecture patterns, Harness streamlines the process of building and managing agent teams. Key features include agent team design, skill generation, orchestration, and validation.

To use Harness, simply trigger it with a prompt like "Build a harness for this project" and it will automatically generate the necessary agent definitions and skills tailored to your domain. The plugin is structured with a plugin.json manifest, SKILL.md definitions, and references for skill authoring and testing guides.

Technical highlights include the use of

plugin marketplace add

for installation and

plugin install harness@harness-marketplace

for setup. The plugin also supports multiple execution modes, including Agent Teams and Subagents, allowing for flexibility in deployment.

Harness is designed for a wide range of audiences, from developers and researchers to business users and educators. Whether you're looking to build a complex software system or simply automate a workflow, Harness provides a powerful tool for creating and managing agent teams.

In a nutshell: Harness revolutionizes team-architecture creation, making it faster and more efficient than ever before!

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

💡 godotengine/godot just hit the trending charts — here's why it matters.

🔗 https://github.com/godotengine/godot
📝 Godot Engine – Multi-platform 2D and 3D game engine
──────────────────────────────

The Godot Engine is a feature-packed, cross-platform game engine for creating 2D and 3D games. It provides a comprehensive set of tools for game development, including a unified interface, and allows for one-click exporting to various platforms like desktop, mobile, and web.

The engine is completely free and open source under the MIT license, giving users full control over their games. Godot is community-driven, with a strong focus on community involvement and contribution.

To get started, users can download official binaries or compile from source. The engine has a large community, with various community channels and resources available, including official documentation and demos.

Here's a simple example of how to create a scene in Godot using GDScript:

extends Node
func _ready():
    print("Hello, World!")

Overall, Godot Engine is an excellent choice for game developers, and its community-driven approach makes it an exciting project to be a part of.
Godot Engine: empowering game developers to create without limits - join the community and start creating today!

──────────────────────────────
🧠 Channel: https://t.me/GithubRe