Skip to content

quickstart-langsmith.md

Latest commit

 

History

History
124 lines (86 loc) · 2.82 KB

quickstart-langsmith.md

File metadata and controls

124 lines (86 loc) · 2.82 KB
Raw

Connect LangSmith in 5 minutes

Pull your LangSmith traces into Sentinel and see your first AI agent insights.

Prerequisites

1. Start the Sentinel stack

git clone https://github.com/sentinelrca/sentinel
cd sentinel/code

task up                                                          # Postgres + ClickHouse + Redis
cd infra/migrations/postgres && uv run alembic upgrade head     # run DB migrations
cd ../../

Start the API and worker in two terminals:

# Terminal 1 — API
cd services/api
uv sync
SENTINEL_SECRET_KEY=$(python3 -c "from cryptography.fernet import Fernet; print(Fernet.generate_key().decode())") \
SENTINEL_ADMIN_KEY=$(openssl rand -hex 32) \
uv run uvicorn sentinel_api.main:app --port 8000
# Terminal 2 — Worker
cd services/worker
uv sync
SENTINEL_SECRET_KEY=<same key as above> \
uv run celery -A sentinel_worker.main worker --loglevel=info

2. Get your LangSmith API key

In LangSmith: Settings → API Keys → Create API Key.

Your key looks like lsv2_pt_....

3. Create your workspace and get an API key

curl -X POST http://localhost:8000/v1/workspaces \
  -H "Content-Type: application/json" \
  -H "X-Admin-Key: $SENTINEL_ADMIN_KEY" \
  -d '{"name": "my-workspace"}'

Save the api_key — it's shown only once. Use it as $API_KEY below.

4. Connect your LangSmith source

curl -X POST http://localhost:8000/v1/sources \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "kind": "langsmith",
    "config_json": {
      "api_key": "lsv2_pt_YOUR_KEY",
      "project_name": "default"
    }
  }'

project_name is the LangSmith project you want to analyze. Defaults to "default" if omitted.

5. Sync your traces

SOURCE_ID="src-..."   # from the response above

curl -X POST http://localhost:8000/v1/sources/$SOURCE_ID/sync \
  -H "Authorization: Bearer $API_KEY"

6. View your insights

curl http://localhost:8000/v1/insights \
  -H "Authorization: Bearer $API_KEY" | python3 -m json.tool

Or open the web UI at http://localhost:3001 (see step 1 of the Langfuse quickstart for UI setup).

LangChain tracing tip

To get traces into LangSmith automatically, add this to your app:

import os
os.environ["LANGCHAIN_TRACING_V2"] = "true"
os.environ["LANGCHAIN_API_KEY"] = "lsv2_pt_..."
os.environ["LANGCHAIN_PROJECT"] = "my-project"

Then run your LangChain/LangGraph app normally — traces appear in LangSmith and Sentinel can pull them.

What Sentinel detects

See the Langfuse quickstart for the full detector list.