Hermit is designed to solve the problem of losing years of context when switching between large language models. It processes your full ChatGPT conversations.json export through a multi-model Anthropic pipeline to preserve the valuable knowledge and context accumulated over time.

The tool generates several key outputs including a global behavioral profile capturing who you are and how you work, a 6-month current snapshot focusing on active priorities rather than ancient history, and up to 25 topic deep-dives covering each ChatGPT project and recurring conversation theme. It creates ready-to-paste files compatible with Claude Memory Import (utilizing ~47% of the 75K capacity), 25 enriched memory bullets for Memory Edits, and output files compatible with Claude Projects, Gemini Gems, and any LLM that accepts system prompts.

Hermit addresses the critical issue of temporal awareness that LLMs lack by calculating recent activity for each theme and applying ACTIVE/PAST labels. This prevents outdated context like "User works at Notion" when you left a year ago or "User is apartment hunting" when you moved in 8 months ago from being carried forward.

The free tier provides comprehensive analytics on your ChatGPT history including conversation count, token statistics, topic clustering, and ChatGPT project detection. The full profiles are available for €19.90 as a one-time payment rather than a subscription model.

Hermit is built for users who want to switch between LLMs without losing their accumulated context, particularly those with extensive ChatGPT histories containing projects, preferences, working styles, and personal context accumulated over years of use.