Milimo Ai Pipeline Expert logo

Milimo Ai Pipeline Expert

Community
mainza-ai
milimo-ai-pipeline-expert

Deep expertise on LTX-2 video pipelines, Flux 2 image/inpainting pipelines, and memory coordination on unified Apple Silicon hardware. Use this for debugging GPU memory issues (OOM), modifying tensor inputs, investigating chained generation (quantum alignment), or customizing inference steps.

Overview

Publishermainza-ai
Repositorymilimovideo
Skill namemilimo-ai-pipeline-expert
Stars
87
Forks
19
Bundled files
Instructions only
Links
Skill sourceRepository
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by mainza-ai on GitHub. Read the source before you install it.

Installation

Install the Milimo Ai Pipeline Expert AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/mainza-ai/milimovideo.git /tmp/milimovideo
mkdir -p .claude/skills
cp -r /tmp/milimovideo/skills/skills/milimo-ai-pipeline-expert .claude/skills/milimo-ai-pipeline-expert
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Milimo Ai Pipeline Expert in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Milimo Ai Pipeline Expert on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Milimo Ai Pipeline Expert is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Milimo AI Pipeline Expert Skill

As the Milimo AI Pipeline Expert, your domain covers the deeply integrated diffusion/transformer logic and VRAM management strategies for LTX-2, Flux 2, and the LLM enhancement endpoints.

Device & Memory Constraints (CRITICAL)

  • Apple Silicon (MPS) is the primary assumed target environment, though CUDA is supported.
  • The MemoryManager (memory_manager.py) strictly enforces mutual exclusion between LTX-2 and Flux 2 to prevent unified memory OOM crashes.
  • Before loading LTX-2, you must call memory_manager.prepare_for("video").
  • Before loading Flux 2, you must call memory_manager.prepare_for("image").
  • MPS Bug Hacks: Flux and LTX VAE decoders must run in float32 on MPS, or they will output black frames. torch.mps.empty_cache() and gc.collect() must be manually executed between pipeline swaps.

LTX-2 Video Pipeline (model_engine.py & tasks/video.py)

  • LTX-2 is a 19B Dual-Stream Transformer. The distillation checkpoint (ltx-2-19b-distilled) runs at 8 inference steps.
  • Three Pipeline Types:
    1. ti2vid: Default Text/Image-to-Video. Replaces noisy latents at frame 0 with input image conditioning.
    2. ic_lora: Used for subject consistency.
    3. keyframe: Used for start-to-end frame interpolation.
  • Prompt Enhancement: By default, user prompts are piped through llm.py before hitting the text encoder.

Chained Generation (tasks/chained.py)

  • Standard LTX-2 supports max 505 frames.
  • Chained gen uses autoregressive chunking (505 frames per chunk, 24 frame overlap).
  • Quantum Alignment: Latent space is 8 pixels per token. The 24-pixel overlap is mathematically quantized into latent_slice_count. Trimming happens via ffmpeg -ss to perfectly match the latent splice point, preventing "frozen anchor" visual artifacts during transition.

Flux 2 Image & Inpainting (models/flux_wrapper.py)

  • FluxInpainter Singleton: Holds the Flow model, AE, Qwen 3 text encoder, and CLIP ViT-L IP-Adapter.
  • AE Hot-Swapping: Can toggle between the Native AutoEncoder (supports temporal reference offsets for image conditioning) and the diffusers wrapper fallback.
  • Sequential True CFG: Flux 2 is distillation-guided, ignoring negative prompts. Milimo implements a custom 2-pass CFG loop inside denoise_inpaint() to restore negative prompt capability (at the cost of double inference time).
  • Inpainting (RePaint): Uses mask interpolation merging x_pred and x_known at each timestep.

Ollama Coordination (llm.py)

  • Prompts enhanced via local VLM running alongside generation.
  • KEEP_ALIVE Management: To prevent VRAM lockout, llm.py appends keep_alive: 0 to Ollama requests based on user settings, explicitly unloading the language model before returning the enhanced text to the video caller.

Frequently asked questions

What does the Milimo Ai Pipeline Expert AI skill do?

Deep expertise on LTX-2 video pipelines, Flux 2 image/inpainting pipelines, and memory coordination on unified Apple Silicon hardware. Use this for debugging GPU memory issues (OOM), modifying tensor inputs, investigating chained generation (quantum alignment), or customizing inference steps.

Why use Milimo Ai Pipeline Expert on TypingMind?

Because you install it once and use it with any model. Milimo Ai Pipeline Expert is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Milimo Ai Pipeline Expert in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/mainza-ai/milimovideo/tree/main/skills/skills/milimo-ai-pipeline-expert. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Milimo Ai Pipeline Expert?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Milimo Ai Pipeline Expert?

As many as you like. As long as a model supports skills, you can use Milimo Ai Pipeline Expert with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Milimo Ai Pipeline Expert AI skill free?

It is published on GitHub by mainza-ai. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

Related AI Skills

View all
affaan-m logo

Database Migrations

affaan-m
CommunityPopular

Buenas prácticas de migración de base de datos para cambios de esquema, migraciones de datos, rollbacks y despliegues de tiempo cero en PostgreSQL, MySQL y ORMs comunes (Prisma, Drizzle, Kysely,…

39.1K
261.1K
Instructions
View
affaan-m logo

Deployment Patterns

affaan-m
CommunityPopular

Flujos de trabajo de despliegue, patrones de pipeline CI/CD, contenedorización Docker, health checks, estrategias de rollback y listas de verificación de preparación para producción de aplicaciones…

39.1K
261.1K
Instructions
View
affaan-m logo

Django Patterns

affaan-m
CommunityPopular

Patrones de arquitectura Django, diseño de API REST con DRF, buenas prácticas de ORM, caché, señales, middleware y aplicaciones Django de nivel producción.

39.1K
261.1K
Instructions
View
affaan-m logo

Docker Patterns

affaan-m
CommunityPopular

Patrones de Docker y Docker Compose para desarrollo local, seguridad de contenedores, networking, estrategias de volúmenes y orquestación de múltiples servicios.

39.1K
261.1K
Instructions
View

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇