v2026.2.13

virattt/dexterv2026.2.13Feb 13, 2026by virattt

AI Summary

This major release introduces WhatsApp gateway support, adds Perplexity as a web search provider, includes a web_fetch tool, and performs significant architectural refactoring with centralized LLM provider registry.

Key Highlights

  • Add WhatsApp gateway with QR linking and self-chat
  • Add Perplexity as optional web search provider
  • Add web_fetch tool for fetching and summarizing web content
  • Refactor agent into run context and tool executor modules
  • Centralize LLM provider registry into single source of truth

New Features

  • WhatsApp gateway with QR linking, access control, routing, and session store
  • Perplexity as optional web search provider
  • web_fetch tool for fetching and summarizing web content
  • Agent refactoring into run context and tool executor modules
  • Centralized LLM provider registry
  • Improved model selector with metadata only for longer tasks
  • WhatsApp reply markdown cleanup (bold formatting)
  • Perplexity max_tokens increased to 4096
  • Fix for issue #100

Full Release Notes

### Changes

- Add WhatsApp gateway: link via QR, self-chat with Dexter; access control, routing, and session store
- Run WhatsApp gateway under Node (tsx) for Baileys stability; handle history catch-up messages (mark read, skip reply)
- Add Perplexity as optional web search provider
- Add `web_fetch` tool for fetching and summarizing web content
- Refactor agent into run context and tool executor modules
- Centralize LLM provider registry into single source of truth
- Improve model selector and show metadata only for longer tasks
- Clean up WhatsApp reply markdown (bold formatting)

### Fixes

- Fix issue #100
- Increase Perplexity max_tokens to 4096