drost-ai/tinker-ai-skill: Fine-tune agentic models on your own coding sessions with Tinker

This open-source repository provides a nine-step workflow and Python scripts for turning coding-agent sessions into fine-tuning datasets and training LoRAs on Tinker. The skill automates dataset collection, cleaning, anonymization, and compaction to produce quantized GGUF models running locally via llama.cpp. Created by Migel Tissera, the pipeline is designed for agents like Claude Code, Codex, and OpenCode. Open article →