What problem does it solve?
Importing your context from another AI into Runneth is slow and risky because it requires reorganizing memory, resolving duplicates, and deciding what actually conflicts with what you already have saved.
Core Features & Use Cases
- Provider-specific handoff prompts: Guides you through extracting your saved memories, instructions, projects, and preferences from ChatGPT, Claude, or Gemini.
- Normalization, deduping, and truncation recovery: Parses the exported JSON/ZIP, validates against a canonical manifest schema, deduplicates by content hash, and triggers a continuation workflow when exports are truncated.
- Inline triage with conflict-aware approval: Classifies each item as behavioral vs contextual, flags conflicts against your existing brain, renders a read-only HTML review, and waits for explicit chat approval (including category selection, ranges, skip, and raise candidates).
- Safe writing with an explicit contract: Writes approved behavioral preferences into your profile so they load every session, writes contextual items into imports for INDEX-based retrieval, and only writes org/user.md raises when you explicitly request and confirm them.
Quick Start
Say to Runneth: Import my context from ChatGPT, then follow the handed-over ChatGPT extraction prompt, upload the resulting export, review the read-only HTML, and approve what you want saved by replying with the corresponding approval command (such as approve all).
