# LangSmith Traced OpenAI Agent

The LangSmith Traced OpenAI Agent MCP server runs a LangChain prompt chain through Gemini on Vertex AI and can send optional traces to LangSmith.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · 4178 installs
- Author: a2a
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://smithery.ai/server/a2a/llm-orchestration-agent
- Updated: 2026-05-31
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/mcp/langsmith-traced-openai-agent

## What it does

The server combines `langchain-core`, `langchain-openai`, `langsmith`, and `openai`. Its `invoke` tool sends a request through a chain with system and human messages, using Gemini on Vertex AI. LangSmith tracing is optional.

The `selftest` tool runs a fixed check at no cost and returns the actual result, an assertion grade, and a trust manifest. That manifest covers a signed Agent Card, JWKS, the governance and licence chain, the safety posture, and a proof link for each claim. Requests go to a remote endpoint, so both tools depend on that endpoint being reachable.

## Requirements
- Remote server: `https://llm-orchestration-agent--a2a.run.tools`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http llm-orchestration-agent https://llm-orchestration-agent--a2a.run.tools
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add langsmith-traced-openai-agent --url https://llm-orchestration-agent--a2a.run.tools
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "langsmith-traced-openai-agent": {
            "url": "https://llm-orchestration-agent--a2a.run.tools"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "langsmith-traced-openai-agent": {
            "type": "http",
            "url": "https://llm-orchestration-agent--a2a.run.tools"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http langsmith-traced-openai-agent https://llm-orchestration-agent--a2a.run.tools
```

## When to use it
- run a prompt through a LangChain chain
- connect Gemini on Vertex AI to an MCP client
- send optional traces to LangSmith
- check the agent with `selftest`
- inspect a trust manifest with claim proofs

## Example prompts
- Run this request through `invoke` with a system instruction and a human message.
- Use `selftest` and return the test result together with the trust manifest.
- Call `invoke` for this question and enable LangSmith tracing.

## Similar
- [Secured LangChain OpenAI Agent](https://affpapa.org/en/nevibecode/mcp/secured-langchain-openai-agent.md) — MCP server · 2894 installs
- [langsmith-fetch](https://affpapa.org/en/nevibecode/skills/langsmith-fetch.md) — skill · 1300 installs
- [langsmith](https://affpapa.org/en/nevibecode/skills/langsmith.md) — skill · 328 installs
- [langsmith-online-eval-engineering](https://affpapa.org/en/nevibecode/skills/langsmith-online-eval-engineering.md) — skill · 409 installs
- [langsmith-trace-analyzer](https://affpapa.org/en/nevibecode/skills/langsmith-trace-analyzer.md) — skill · ★ 106 repo stars
- [langsmith-tracing](https://affpapa.org/en/nevibecode/hooks/langsmith-tracing.md) — hook · 204 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
