News for hackers. Stuff that matters.

shivnathtathe

Show HN: MCP bridge for chat between OpenCode and Claude Code sessions

https://github.com/shivnathtathe/openbridge

I didn’t want to copy and paste messages between coding-agent sessions or across vendors, so I built an MCP server that lets one model in one session talk to another model in a different session. They can even have a group chat. It supports multiple OpenCode sessions and lets GPT-6.1 in OpenCode talk to Claude in Claude Code.

( Read More... | no comments | on HN )
yes! opensmith export dumps to JSON out of the box. Parquet isn't there yet but it's a good shout — would just need a pandas dependency. if you open an issue I'll prioritize it.
Hey HN. I built this because LangSmith needs a cloud account to see my own traces.

The full stack is: → @trace decorator on any Python function → autopatch() for OpenAI, Anthropic, LiteLLM, Qdrant, ChromaDB, Pinecone → SQLite at ~/.opensmith/traces.db → Dashboard at localhost:7823

800+ downloads in 2 days.

I have not done any marketing.

Its early in development. I am happy to hear what is missing.

Show HN: I built an open-source, local-first alternative to LangSmith

https://pypi.org/project/opensmith/

Hey HN. Built this because LangSmith needs a cloud account just to see my own traces.pip install opensmith — everything shows up at localhost:7823. SQLite. No accounts.No config. Fully offline.800+ downloads in 2 days. Surprised me.Would love honest feedback from this community.

( Read More... | 1 comment | 5 points | on HN )
Interesting approach! I've been working on something complementary — local-first LLM observability called opensmith. Different angle but same philosophy of keeping everything local.
Re:Agents need control flow, not more prompts on Friday May 08, 2026 @05:17AM (#48060648)
If you have any org email, you can get free chatgpt + subscription.
Re:Agents need control flow, not more prompts on Friday May 08, 2026 @05:17AM (#48060641)
Observability is the missing piece here — built opensmith for exactly this reason, tracing agent control flow locally
Re:DeepSeek 4 Flash local inference engine for Metal on Friday May 08, 2026 @05:16AM (#48060636)
Been working on local-first LLM observability for exactly this use case — tracing local model pipelines without sending data to cloud. Happy to share if anyone's interested.