Skill UI

This guide provides a comprehensive tutorial on running various AI inference tasks using the Together AI API, which is compatible with the OpenAI standard. It demonstrates key functionalities including chat completions, streaming responses, image generation, and creating embeddings. This is ideal for developers who need to test open-source models, compare performance across different LLMs (like Llama or Mixtral), or integrate powerful generative AI features into their applications using Python or Node.js.

This guide provides comprehensive instructions for setting up and authenticating the Together AI SDK. It covers installation using both Python and Node.js, configuring the necessary API key, and verifying connectivity to utilize the OpenAI-compatible API for powerful open-source model inference and fine-tuning tasks. Ideal for developers starting with Together AI.

The Ejentum Reasoning Harness is a comprehensive library of cognitive operations designed to guide complex AI agents through critical thinking tasks. It exposes four specialized harnesses (reasoning, code, anti-deception, memory) that mitigate common failure modes like hallucination and reasoning decay. Instead of a fixed answer, the agent ingests a detailed, executable scaffold—including failure patterns, procedures, and falsification tests—to ensure robust, self-verified output.

Open Dynamic Workflows (ODW) is an advanced, open-source engine designed to manage complex coding tasks using multiple AI agents. It operates in three stages: planning the task into parallel subtasks, orchestrating various coding agents (like Codex or Antigravity) to work concurrently, and finally, performing an adversarial verification pass to challenge and validate the combined output before merging. It supports multiple model providers and is ideal for structured, high-reliability software development.

SkillOpt-Sleep provides a mechanism for autonomous agent self-improvement. By simulating a 'sleep' or 'dream' cycle, it harvests past interaction sessions, mines recurring tasks, and replays them through the designated backend. It consolidates learned knowledge into persistent memory and skills, ensuring that only strictly validated improvements are adopted via a held-out gate, simulating long-term competence gain from short-term experience. Ideal for scheduled, background optimization.

This skill provides a comprehensive workflow for generating, editing, and manipulating images using Google's Gemini Nano Banana Pro model (gemini-3-pro-image-preview). It is designed for tasks such as creating visuals, generating logos, modifying photos, or creating product mockups, offering a robust, file-based approach to handle complex image data and API interactions.

This skill provides utilities to manage and drain the deliberation queue (brain_frames) of the ACC system. It allows users to list waiting frames, retrieve necessary context, perform deliberation on open tasks, and submit continuations using `acc_act(runtime="continue")`. It is essential for checkpointed work saved by headless runs, ensuring continuous task resolution and calibration against predicted outcomes in complex AI workflows.

This professional tool helps developers and researchers identify the optimal AI model for a specific task, such as coding, reasoning, or image recognition. It queries official HuggingFace leaderboards, filters models based on hardware constraints (e.g., available VRAM or RAM), and provides a detailed comparison table including benchmark scores, parameter size, and licensing information.

This skill is designed for complex, multi-step tasks requiring coordination among multiple AI agents. It implements a robust workflow engine that first generates a detailed plan (task graph, topology) and then orchestrates parallel execution with adversarial verification. Use this when the task spans many files, requires multiple specialized roles, or involves an overall 'workflow' concept, ensuring structured, reliable, and auditable results.

A comprehensive guide for training or fine-tuning advanced sentence-transformer models. This skill assists in selecting the correct architecture—including Bi-encoders (SentenceTransformer), CrossEncoders (for reranking), and SparseEncoders—for various Natural Language Processing (NLP) tasks. Use cases include semantic retrieval, similarity matching, document clustering, and pair-wise classification.

This skill provides expert capabilities for training and fine-tuning transformer language models using the TRL (Transformers Reinforcement Learning) library. It supports state-of-the-art post-training methods, including Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), Group Relative Policy Optimization (GRPO), and Reward Model training. Use this skill to align and customize foundation models for advanced tasks via CLI commands.

A powerful plugin enabling Claude Code to delegate complex tasks to external AI CLIs (such as OpenAI Codex or Google Antigravity/Gemini) without exiting the conversation. It supports multi-model workflows, allowing users to seek a second opinion, cross-check deep analysis, or delegate side tasks (like research or code review). The core value is that Claude synthesizes the raw outputs, providing a critical review and reconciliation of findings across all consulted models.