Skip to content

feat(chatgpt-local): add parser-based conversation scanner - #676

Open
Lefeaker wants to merge 2 commits into
ferdium:mainfrom
Lefeaker:feat/chatgpt-message-extract
Open

feat(chatgpt-local): add parser-based conversation scanner#676
Lefeaker wants to merge 2 commits into
ferdium:mainfrom
Lefeaker:feat/chatgpt-message-extract

Conversation

@Lefeaker

@Lefeaker Lefeaker commented Mar 7, 2026

Copy link
Copy Markdown

Summary

Add a new chatgpt-local recipe spike with a parser-based conversation scanner.

This vendors a multi-platform parser strategy inspired by AllinOB, then uses the ChatGPT parser in webview.js to scan the current conversation DOM, normalize messages, and send scan results back to the host app.

What Changed

  • add recipes/chatgpt-local
  • vendor parser modules for multiple AI platforms
    • ChatGPT
    • Claude
    • Copilot
    • DeepSeek
    • Gemini
    • Kimi
    • Tongyi
  • extract shared parser utilities
    • DOM helpers
    • markdown conversion
    • parser core / registry
  • add chatgpt-local/webview.js
    • scans current conversation
    • exposes window.__AI_HUB_RESCAN_CURRENT_CONVERSATION()
    • auto rescans on URL change
    • sends aihub-scan-messages payloads to host

Why

This is the recipe-side foundation for the local archive MVP. It keeps parser logic inside the recipe runtime while making the app-side archive pipeline platform-agnostic.

Validation

  • packaged recipes successfully
  • verified webview console output for ChatGPT conversation scans
  • verified manual rescan bridge works from DevTools
  • verified URL-change auto rescan picks up conversation switches

Scope Notes

  • only ChatGPT Local is wired into the host archive flow in this phase
  • other vendored platform parsers are included as strategy/reference groundwork, not yet host-wired

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant