

Model Context Protocol (MCP) and Ollama-Powered Cloud-Local LLM Hybrid Inference Orchestrator
Automatically translated from Japanese by AI

MCP server inspection and comparison tool. With tool definition quality evaluation and security scanning features.

Give your AI persistent memory and a workspace.

An MCP server that sends rich notifications through OS desktop notifications and Discord Webhooks. Compatible with various AI agents such as Gemini CLI, Claude Code, Codex CLI, etc.

This is a Mermaid Editor-type application that can directly convert Mermaid into flowcharts and mind maps. Compared to existing editors, it allows you to freely customize the appearance and styling. As a trial, we've equipped it with Google's Built-in AI functionality, so you can output code in Mermaid format within the app. However, since this requires various setup procedures, if it's too much trouble, you can simply paste code that you've generated from ChatGPT, Gemini, or similar services directly into the app.

AI Knowledge Fusion Device Collect, create, correct, dig deeper, manage, listen with your ears openManidoc becomes an excellent app depending on your ideas

This is a translation application using a local LLM. You select a GGUF file, choose the source and target languages, input text, and press the Translate button. If the model supports those languages, it will translate for you. The application uses Rust and Python—Rust handles the main processing and GUI, while Python handles LLM-related operations. Of course, you can change various parameters like temperature and topp/k, as well as the system prompt from the app side. You can add translation languages and configure the model and app settings by editing config.yaml. If you don't add fonts, the translation will display as □□□ and so on, so please place TTF or OTF files in the fonts folder and set the language and font file paths in fonts.yaml. fonts.yaml example Additionally, as a bonus feature, it uses the TTS (Text to Speech) model Kokoro-82M. This enables voice generation for some languages including Japanese and English. (For long text, I handled it by splitting by tokens and concatenating them afterward.) For languages not supported by Kokoro-82M, while pre-downloading is required, voice generation uses the OS standard text-to-speech functionality. I'm not considering support for anything other than Kokoro-82M. Since voice generation is just a bonus feature, please don't expect too much from it. ※The name "3LM" is just an abbreviation of LocalLLM. Since it's LLM, it becomes 3LM.

An AI meeting minutes and translation app that works without an internet connection.

A tool to view Markdown files beautifully

A tmux-like application that can split the screen on a terminal, launch multiple Claude Code instances, send messages, and launch Claude Code. Multi-platform compatible.

AA (ASCII art) is convenient when conveying UI images to LLMs (AI). While you could try to convey images visually, that only works with multimodal-capable models. However, with AA, even LLMs with very low parameters can read and understand it. You can convey UI images without being limited by the model type. Plus, with AA, you can make edits in a text editor. No specialized tools are needed. However, creating AA is difficult. That's why I created a tool that makes it easy to generate AA by arranging parts in a PowerPoint-like interface. It supports .md, .txt, and .png output. It has layer separation functionality, real-time preview, and the ability to load and reference images while creating, so I believe it has sufficient performance for practical use.

An app for local captioning and translation of multinational team meeting audio on Mac

The Prompt BAR is a Windows tool that converts images written in Japanese into English prompts for image generation.

フリーランスとクライアントのファイル共有・進捗管理を、1つのURLで完結するポータルサービス。

An App That Eliminates "I Don't Know What to Ask ChatGPT" Choose from 1,033 prompt templates, fill in the dropdowns, and get professional-quality questions in 30 seconds. One-tap paste to ChatGPT / Gemini / Claude / Perplexity / Genspark. --- Why I Built It I wanted people to use AI as a tool for solving small daily frustrations, and as a prescription for reclaiming lost time. Proofreading English emails, deciding tonight's dinner, praising children, getting a second opinion on investments. People are losing a little time and energy every day on small tasks that AI should be able to handle instantly. --- Frustrations with Existing AI Prompt Apps It's incredibly tedious to find a prompt you like. Read the prompt → rewrite it for yourself → copy and paste—most people drop out at these 3 steps. In the end, people just throw vague requests like "make this sound nice" and cry over mediocre answers. This App's Approach Just follow the basic template and tap to select the parts that differ for each user as "variables". No rewriting needed—you get a prompt perfectly optimized for your situation. For example, if you want to know "how to praise my child": → Tap the age (3 years old / early elementary / middle school...) → Tap the scene (extracurricular activities / studying / chores...) → Tap your parenting style (praise to encourage growth / strict approach...) That's it—you get a prompt that fits your situation perfectly. --- What I'm Aiming For Ultimately, I want this to become a "beginner's guide" to AI chat apps. People who installed ChatGPT but never figured out how to use it can use this app as a stepping stone to incorporate AI into their daily lives. That's the bridge-building role I'm creating this for. --- Specs Number of prompts: 1,033 (17 categories / 148 subcategories) Work / cooking / health / investing / parenting / learning / job hunting / SNS / fortune telling, etc. Login bonuses / streaks / referral codes / 14-day free premium Freemium: 51 prompts available for free Tech Stack Expo (React Native) + TypeScript Supabase (analytics / rankings / referral code management) RevenueCat (subscription management) expo-notifications (streak warnings / update reminders via local notifications) --- Since this is an indie project, all feedback, bug reports, and feature requests come through loud and clear. If it sounds like your thing, definitely give it a try.

もう1からプロンプトを作らない

You can manage prompts in a list.

Intuitively reproduce brainstorming idea organization from memo writing with a post-it style UI. Output idea organization as JSON and pass the baton to AI concisely.

Excelの工程管理、そろそろ卒業しませんか?

1. Just ask AI, and work reaches remote workers. New task matching 2. Your "request" becomes someone's job. Task matching for remote workers 3. Leave ordering to AI. Remote workers and work

A draft support tool that completes your request to Claude just by filling in items. One-click copying from template to completion.

A draft support tool where simply filling in items completes your request to Claude. One-click copying from template to draft.

This is a free schedule coordination service that works with LINE and Discord. Simply create candidate dates and share the URL, and participants can easily respond with ○ or △. It also supports anonymous participation, so you can start coordinating schedules right away without registering as a member. The organizer can manage everything in one place: creating candidate dates filtered by day of the week, setting deadlines, setting minimum and maximum number of participants, and sharing the meeting location. Reminders to non-respondents and confirmation notifications can be automated through LINE and Discord integration. Once enough participants are gathered, a proposed date is presented, and you can proceed directly to confirming the schedule. It's ideal for people who want to finish coordinating schedules for drinking parties, meetups, club activities, and plans with friends with as little hassle as possible.

Task management app for parallel careers

This is an MCP server that enables cloud LLMs to treat local LLMs running through Ollama as pseudo sub-agents. It aims to improve accuracy and token efficiency at the cost of time.
What it essentially does is allow cloud LLMs to perform ```bash
ollama run <モデル> "なんかいい感じのPythonコード書いて"
The models used can be configured via a TOML file. When doing so, make sure the model names match those that appear in `ollama list`. Role and parameter settings are also configurable.
Since it's an MCP server, theoretically, if it fits in memory, all operations can be performed with local LLMs.
The workflow is as follows:
1. The cloud LLM considers which model and prompt to use, then makes a call
2. The implementation role writes out the code
3. The review role checks it
4. Return to the cloud LLM, and if it's usable, use it as-is; if there are issues, make corrections (sometimes returning to step 2)
Even just creating this kind of flow should change token efficiency.
Generating from 0 to 1 consumes quite a lot of tokens, but by pre-generating from 0 to 0.5 and having the cloud LLM side focus on corrections and refinements, we can improve token efficiency somewhat. However, there is the drawback that inference can take a very long time depending on the environment, so careful judgment about when to use it is necessary.
※The name comes from cloud → sky (空) and local → region (地), creating "Tenchi MCP" (天地MCP - Heaven and Earth MCP).
Feedback
Share your thoughts, bug reports, or suggestions directly with the developer
Log in to share your feedback
Log in to leave feedback