Mpromp2 — the AI Prompt GeneratorThemeClassic ModernDarkLightSlate

Skills Library / Qwen & Kimi

Qwen & Kimi Long-Context Project Setup

Qwen (Alibaba) and Kimi (Moonshot AI) are built around very long context: whole document sets, codebases, and book-length material in a single session. Their skills layer is the project setup - how you load and instruct that material. This page gives you the working pattern and a setup prompt to copy.

How it works

Neither model needs a skills file to shine; they need a discipline: load the corpus first, instruct the model to treat it as the authority, and require it to distinguish what the corpus says from what it knows. The reusable asset is a setup prompt - a header you paste at the start of a session that establishes the corpus rules - plus a reference sheet convention inside the material itself. This mirrors the CLAUDE.md pattern: standing instructions, stated once, governing everything after.

Setup, step by step

1. Assemble the corpus as clean text files; name them descriptively - the model cites by name
2. Open the session with the setup prompt below, pasted before any material
3. Load the material in labeled blocks, one document per block
4. Ask for a coverage map first: what the corpus contains, in outline - verify it read everything
5. Only then start tasks; require source labels on claims drawn from the corpus

Copy-ready starter file

Field notes

The coverage-map step is the acceptance check for long-context work: if the model cannot outline the whole corpus, it has not actually loaded it, and everything after is suspect.

Keep going

Mpromp2 — the AI Prompt Generator. Every guide and skill file here is free to use, including commercially.