Develop Python-based Boost modules for Harbor's LLM proxy pipeline.
npx skills add https://github.com/av/harbor --skill new-boost-module
Please help me install this Agent Skill. Skill: new-boost-module Source: https://github.com/av/harbor/tree/main/.agents/skills/new-boost-module Command: npx skills add https://github.com/av/harbor --skill new-boost-module
The Skill addresses the challenge of creating and utilizing custom Boost modules to enhance Harbor's LLM proxy capabilities. This includes building custom plugins, adding features to the proxy pipeline, and transforming LLM chat completions.
Write a Python module with ID_PREFIX = "my_module" and implement the apply(chat, llm) function to add a new Boost module for Harbor.
High-intent search queries and answers about installing and using this skill.
The Harbor Boost architecture supports custom Python modules to extend the LLM proxy pipeline, enabling the integration of reasoning chains, prompt engineering, and output transforms directly into chat completion processing.
You can add custom middleware to an LLM proxy pipeline by developing a Python-based Boost module that implements the `apply(chat, llm)` function, inserting custom logic directly into the Harbor proxy's chat completion processing flow.
Familiarity with Python and the Harbor Boost architecture is required to effectively develop, manage, and integrate custom proxy modules for transforming LLM chat completions.
Yes, you can transform LLM chat completions by building a custom Python Boost module that implements the `apply(chat, llm)` function, allowing you to intercept and modify chat data within the Harbor proxy pipeline.
The best way to integrate prompt engineering into a proxy layer is using custom Python Boost modules, which provide the necessary templates and framework to inject prompt transformations directly into the LLM chat completion pipeline.
You should avoid building a custom module for LLM proxy processing if your task does not require pipeline-level transformations, reasoning chain integration, or custom middleware, as the standard Harbor proxy setup handles basic chat completions without custom Python extensions.
avDevelop Python modules for custom processing of LLM chat completions in Harbor Boost.
rbkayzManage Open WebUI models, chat completions, RAG uploads, and Ollama operations via API.
GPTtangImplement few-shot learning and chain-of-thought patterns for structured LLM outputs.
pratosExecute Python scripts with uv run and type check with ty check.
