Extract reusable Markdown skills from documents via a multi-agent self-play loop.
npx skills add https://github.com/eluckydog/Ctx2Skill- --skill ctx2skill
Please help me install this Agent Skill. Skill: ctx2skill Source: https://github.com/eluckydog/Ctx2Skill-/tree/main Command: npx skills add https://github.com/eluckydog/Ctx2Skill- --skill ctx2skill
Ctx2Skill automatically discovers, refines, and selects context-specific skills from complex documents using a self-evolving multi-agent loop, eliminating the need for manual annotation or external feedback.
Run the module with your context text and an OpenAI-compatible API, then feed the produced skill set into your next inference task.
High-intent search queries and answers about installing and using this skill.
You can extract reusable skills from complex documents without manual annotation by using a self-play multi-agent loop. This approach autonomously discovers and refines context-specific knowledge through a cross-time replay mechanism, outputting a reusable Markdown skill set.
Self-play multi-agent skill discovery is an unsupervised extraction process using five agent roles—Challenger, Reasoner, Judge, Proposer, and Generator. These roles iteratively interact to extract robust context knowledge that generalizes across tasks without external feedback.
You prepend extracted context skills to system prompts by generating a self-contained Markdown document during training. This skill Markdown is then prepended to your system prompts at inference time, enhancing model performance without modifying underlying models.
Yes, you can use Python APIs to automate context skill extraction. The system provides both command-line and Python API access, allowing easy integration into existing LLM workflows to autonomously produce skill sets from your context text.
No, this multi-agent skill extraction requires neither external feedback nor modifications to underlying models. It uses a self-evolving loop and cross-time replay to independently refine skills, enabling zero or minimal downtime deployment via prepended Markdown.
The best way to identify robust skills that generalize across tasks is using a cross-time replay mechanism within a self-play loop. This process ensures that the discovered context-specific skills remain effective across different scenarios and timeframes.
lordmacuSummarize long-form text and local UTF-8 files via OpenAI-compatible chat completions.
boshu2Inject prior knowledge from .agents artifacts into session context at SessionStart.
quazfentonManage local Ollama models, generate text, and create embeddings via CLI.
LynxLabVNFine-tune large language models with TRL using SFT, DPO, PPO, GRPO, and reward modeling.
