Saki-tw/why-your-ai-is-dumbing-down
A forensic analysis of how local AI IDEs and wrappers secretly inject 'CHECKPOINT' mechanisms and override prompts to lobotomize your LLMs (Claude/Gemini) for token-saving. It's not the model degrading; it's the infrastructure hijacking your context.
GitHub repository with 8 stars and 0 forks.
Topics: ai-agents, claude, context-window, debugging, llm, observability, prompt-engineering, system-architecture