Skip to content

Automate browser based workflows with AI

[NO-TICKET] Update Laminar Usage (#5322)

Aaron Perez39fc16ae69b4
diff --git a/.env.example b/.env.example
index 8871a2dff..3405d14b4 100644
--- a/.env.example
+++ b/.env.example
@@ -110,6 +110,13 @@ PORT=8000
# ANALYTICS_ID: Distinct analytics ID (a UUID is generated if left blank).
ANALYTICS_ID="anonymous"
+# LAMINAR
+# Skyvern's backend runs on port 8000 by default. Consider updating your self-hosted laminar to env vars to avoid conflicts
+# LMNR_HTTP_PORT=8010
+# LMNR_GRPC_PORT=8011
+# LMNR_BASE_URL=http://localhost
+# LMNR_PROJECT_API_KEY=<your-laminar-project-api-key>
+
# 1Password Integration
# OP_SERVICE_ACCOUNT_TOKEN: API token for 1Password integration
OP_SERVICE_ACCOUNT_TOKEN=""
diff --git a/docs/debugging/observability-with-laminar.mdx b/docs/debugging/observability-with-laminar.mdx
index 1d0e876e5..173682126 100644
--- a/docs/debugging/observability-with-laminar.mdx
+++ b/docs/debugging/observability-with-laminar.mdx
@@ -184,9 +184,18 @@ If you're running Skyvern on your own infrastructure, add these to your server's
LMNR_PROJECT_API_KEY=your-laminar-api-key
```
-Skyvern's server includes a built-in `LaminarTrace` integration that initializes Laminar with the LiteLLM callback, capturing every LLM call, token count, and cost. It disables the automatic Skyvern/Patchright instrumentors (to avoid conflicts) and uses Laminar's `@observe` decorator on internal methods instead.
+Skyvern's server initializes Laminar at startup, which auto-instruments LiteLLM to capture every LLM call, token count, and cost. No manual callback setup is needed.
-No code changes needed — once the env var is set, traces appear in your Laminar project automatically.
+If you're running a self-hosted Laminar instance, also set the base URL and ports:
+
+```bash .env
+LMNR_PROJECT_API_KEY=your-laminar-api-key
+LMNR_BASE_URL=http://localhost
+LMNR_GRPC_PORT=8011
+LMNR_HTTP_PORT=8010
+```
+
+No code changes needed — once the env vars are set, traces appear in your Laminar project automatically.
---
diff --git a/fern/observability/overview.mdx b/fern/observability/overview.mdx
index ebb3dc7ef..b51488404 100644
--- a/fern/observability/overview.mdx
+++ b/fern/observability/overview.mdx
@@ -18,24 +18,18 @@ Copy of this guide is available in the [Laminar documentation](https://docs.lmnr
## Quickstart
-To trace Skyvern workflows with Laminar, **initialize Laminar and configure LiteLLM callbacks at the top of your project**. This will automatically capture all LLM calls, browser session recordings, and workflow execution details.
+To trace Skyvern workflows with Laminar, **initialize Laminar at the top of your project**. This will automatically instrument LiteLLM and capture all LLM calls, browser session recordings, and workflow execution details. No manual callback setup is needed.
-```python {3-4} {8-12} {14-15}
+```python {3} {7-8}
from skyvern import Skyvern
import asyncio
-import litellm
-from lmnr import Laminar, LaminarLiteLLMCallback, Instruments
+from lmnr import Laminar
from dotenv import load_dotenv
load_dotenv()
-# Initialize Laminar
-# This will automatically trace all Skyvern functions
-# Disable OpenAI to avoid double instrumentation of LLM calls
-Laminar.initialize(disabled_instruments=set([Instruments.OPENAI]))
-
-# Configure LiteLLM to trace all LLM calls made by Skyvern
-litellm.callbacks = [LaminarLiteLLMCallback()]
+# Initialize Laminar — automatically instruments LiteLLM
+Laminar.initialize()
skyvern = Skyvern(api_key="YOUR_API_KEY")
@@ -49,6 +43,10 @@ if __name__ == "__main__":
asyncio.run(main())
```
+<Warning>
+`LaminarLiteLLMCallback` and manually setting `litellm.callbacks` are deprecated. `Laminar.initialize()` handles LiteLLM instrumentation automatically.
+</Warning>
+
## Viewing Traces
You can view traces in the Laminar UI by navigating to the traces tab in your project. When you select a trace, you can see:
diff --git a/skyvern/forge/api_app.py b/skyvern/forge/api_app.py
index c09de626d..0cc433f1a 100644
--- a/skyvern/forge/api_app.py
+++ b/skyvern/forge/api_app.py
@@ -1,3 +1,4 @@
+import os
import uuid
from contextlib import asynccontextmanager
from datetime import datetime
@@ -211,6 +212,25 @@ def create_api_app() -> FastAPI:
forge_app_instance = start_forge_app()
+ # Initialize Laminar tracing after ForgeApp so auto-instrumentation works.
+ lmnr_api_key = os.environ.get("LMNR_PROJECT_API_KEY")
+ if lmnr_api_key:
+ try:
+ from lmnr import Laminar # noqa: PLC0415
+
+ lmnr_base_url = os.environ.get("LMNR_BASE_URL", "http://localhost")
+ lmnr_grpc_port = int(os.environ.get("LMNR_GRPC_PORT", "8001"))
+ lmnr_http_port = int(os.environ.get("LMNR_HTTP_PORT", "8000"))
+ Laminar.initialize(
+ project_api_key=lmnr_api_key,
+ base_url=lmnr_base_url,
+ grpc_port=lmnr_grpc_port,
+ http_port=lmnr_http_port,
+ )
+ LOG.info("Laminar tracing initialized", base_url=lmnr_base_url, grpc_port=lmnr_grpc_port)
+ except Exception as e:
+ LOG.warning("Failed to initialize Laminar tracing", error=str(e))
+
fastapi_app = FastAPI(lifespan=lifespan)
# Add CORS middleware