Skip to main content

ChatArchive Development Roadmap

This document outlines the milestones for turning ChatArchive from a ChatGPT viewer into a robust, provider-neutral conversational knowledge base.

1. Provider Adapters (Phase 1)​

Build import adapters that map other platforms into the shared archive model:

  • Claude Export Adapter: Ingest exported JSON from Anthropic/Claude.
  • Gemini Export Adapter: Ingest exported chat logs from Google.
  • Shared Link Parser: Ingest single shared HTML or JSON conversation links.
  • Local Tool Adapters: Support export layouts from local platforms (Jan, Ollama, Open WebUI).

2. Local LLM Continuation (Phase 2)​

Use archived conversations as high-context starter prompts for local models:

  • Markdown Context Exporters: Clean export tools to generate model-friendly context packs.
  • Context Batching & Summary: Split long chats into chronological chunks, generating rolling summaries to fit within local context windows.
  • Ollama Prompts: Package histories directly into Ollama-compatible configurations.

3. Search & Retrieval (Phase 3)​

Move from basic client-side filters to database-level querying:

  • Full-Text SQLite FTS5 Indexing: Weighted search index for titles, messages, and code blocks.
  • Semantic Search: Build local embeddings via native packages to allow query matches by concept.
  • Cross-Conversation Sibling Search: Match related chats based on shared files, topics, or code blocks.

4. Editing & Curation (Phase 4)​

Manage your history like a personal knowledge wiki:

  • Local Renaming & Summarizing: Override provider titles with custom labels.
  • Cross-Provider Collections: Group conversations by topic or project (e.g. create a "Rust Tauri App Setup" collection containing chats from ChatGPT and Claude).
  • Inline Annotations: Add custom notes or highlight key decisions inside a thread.

5. Privacy & Redaction (Phase 5)​

Ensure credentials and secrets are scrubbed before publishing:

  • Scrubbing Engine: Scan conversations for API keys, email addresses, passwords, and custom regex patterns.
  • "Public Build" Filter: Strip system prompts, raw tool calls, or labeled messages from the index before publishing.
  • Redaction Logs: Generate diff reports showing what information was stripped.

6. Static Publishing (Phase 6)​

Host your archives on the web:

  • Pages Integration: Build templates compatible with GitHub Pages and Cloudflare Pages.
  • Single-Page Share: Export a single conversation as a self-contained, offline-compatible HTML file.
  • Password Protection: Add client-side password gates for personal hosted archives.