向助手发出协助请求,适用中华文化理解、中文古文理解、中国特色作品创作、小红书与抖音的宣发文案编写、编写测试用例、信息检索等任务场景,同时也是其他专家助手的强力替补与协助,可以作为大部分认为的备选候补方案。
The 'run-iflow' skill is designed to facilitate requests for assistance from external assistants through the MCP tool. It offers an isolated environment where tasks such as understanding Chinese culture, interpreting classical Chinese literature, creating Chinese-style works, drafting promotional copy for platforms like XiaoHongShu (Little Red Book) and Douyin (TikTok), writing test cases, and information retrieval can be executed. By creating a clean and runtime-isolated environment, it ensures that the requested tasks are carried out with minimal context interference, ensuring high relevance and accuracy in the result. This skill acts as a strong alternative or backup for various expert assistants, especially in cases where specific regional knowledge or highly tailored tasks are required.
The main features of 'run-iflow' include the ability to interact with a range of different models, such as GLM-4.6, Qwen3-Coder-Plus, DeepSeek-V3.2, Kimi-K2-0905, and MiniMax-M2, which specialize in various domains. It allows users to send a fully formulated task description along with optional system instructions, specify a working directory, and choose a model that best fits the task's requirements. Additionally, users can customize environment variables to tailor the task execution further. Use cases for this skill include tasks like content creation for Chinese social media platforms, programming-related tasks, and detailed information retrieval. Target users include developers, content creators, marketers, and individuals requiring specialized knowledge in Chinese culture or specific technical domains.
You can use the 'mcp__plugin_headless-knight_runCLI__iflow' tool to send a request, providing parameters such as 'prompt', 'systemPrompt', and model specifications.
The available models are GLM-4.6, Qwen3-Coder-Plus, DeepSeek-V3.2, Kimi-K2-0905, and MiniMax-M2, each suited for different types of tasks.
Yes, you can customize environment variables by providing an 'env' object with key-value pairs.
You must provide a complete task description in the 'prompt' parameter, and optionally, you can specify a 'systemPrompt' and other parameters such as 'workDir' and 'model'.
If you don't specify a model, 'MiniMax-M2' will be used by default. The working directory defaults to the current directory unless otherwise specified.
Quick Setup:
.claude/skills/Repository
LostAbaddon/HeadlessKnight