Xinwei He
188 posts

Xinwei He
@xinwei_97
Founder and CEO of @TracerootAI
San Francisco, CA Katılım Temmuz 2025
159 Takip Edilen211 Takipçiler

@TracerootAI now has a span timeline view 🚀
Every span as a Gantt-style bar — latency hotspots jump out instantly.
Click a bar → jumps to the span in the tree.
Works on live traces too, bars stream in as the agent runs!
English

@TracerootAI now supports live tracing 🚀
Long-running AI agents are painful to debug — you stare at a blank screen for minutes, then sift through everything at once.
With live tracing, spans now stream into TraceRoot as they happen. You watch tool calls, LLM requests, and errors populate while the agent is still running.
What's new:
- Live span streaming — SSE + Redis pub/sub
- Auto-refresh on the traces list
- Zero SDK changes — works across Claude Agent SDK, OpenAI Agents, Gemini, LangChain, and more
Give it a try!
👉 pip install traceroot
English

@internetDevv @cognition Checked and did not get any message from you.
English

Hey @cognition 👋 The DeepWiki page for our open-source repo is severely outdated and the "refresh" button doesn't work. Support buttons on docs.devin.ai and app.devin.ai are also no-ops.
We'd love to point prospects to deepwiki.com/traceroot-ai/t… instead of sending zip files or hosting our own deepwiki. Can you help?
🔗 github.com/traceroot-ai/t…



English

Thanks! The content feels pretty up-to-date for all pages now, though a nit would be for some pages e.g. deepwiki.com/traceroot-ai/t… even though the content is definitely up-to-date but in the top left corner the `Last Indexed` section is still shows a much earlier date in this case `17 March 2026 (7721b2)` instead of the latest `3 May 2026 (1bf515)` would could be confusing to viewers.

English

@xinwei_97 @cognition Hey @xinwei_97 I just refreshed your repo's DeepWiki! For more frequent and on-demand indexing, as well as access to more powerful Ask, we would love for you to try out Devin. Happy to help get you set up with some free credits, just DM me.
English

@michael_chomsky Request for revision and ask to split the PR into multiple PRs each with < 800 LoC, otherwise closed
English





















