Skip to main content

Quickstart

This guide gets you from zero to your first trace and risk score. You have two ways to start: use Splyntra Cloud (managed, nothing to run) or self-host the open core locally. Both speak the same API.

1. Get an endpoint and API key​

Option A — Splyntra Cloud​

  1. Sign up at app.splyntra.com.
  2. Create a project. Copy the ingest endpoint and an API key from Settings → API keys.

Option B — Self-host​

Run the open core locally:

git clone https://github.com/splyntra/splyntra
cd splyntra
docker compose up

The collector listens on http://localhost:4318 (OTLP/HTTP) and the dashboard on http://localhost:3000. See Self-hosting for details.

note

API keys are stored only as SHA-256 hashes on the server — the raw key is shown once at creation. Keep it secret and set it as an environment variable rather than committing it.

2. Instrument your agent​

Splyntra is OpenTelemetry-native, so instrumentation is one import plus one init call. Pick your framework below.

LangGraph / LangChain​

import os
from splyntra import instrument

instrument(
    endpoint=os.environ["SPLYNTRA_ENDPOINT"],   # e.g. https://ingest.splyntra.com
    api_key=os.environ["SPLYNTRA_API_KEY"],
    project="support-agent",
)

# Build and run your graph as usual — every node, tool, and model call is now traced.
from langgraph.prebuilt import create_react_agent
agent = create_react_agent(model, tools)
agent.invoke({"messages": [("user", "What's our refund policy?")]})

OpenAI Agents SDK​

import os
from splyntra import instrument

instrument(
    endpoint=os.environ["SPLYNTRA_ENDPOINT"],
    api_key=os.environ["SPLYNTRA_API_KEY"],
    project="research-agent",
)

from agents import Agent, Runner
agent = Agent(name="researcher", instructions="Answer with sources.")
Runner.run_sync(agent, "Summarize the latest earnings call.")

CrewAI​

import os
from splyntra import instrument

instrument(
    endpoint=os.environ["SPLYNTRA_ENDPOINT"],
    api_key=os.environ["SPLYNTRA_API_KEY"],
    project="ops-crew",
)

# Your crew and tasks run unchanged; each agent and task becomes a span.
crew.kickoff()

Raw OpenTelemetry (any language)​

Already emitting OTel? Point your exporter at Splyntra's OTLP/HTTP endpoint — no SDK required:

export OTEL_EXPORTER_OTLP_ENDPOINT="https://ingest.splyntra.com"
export OTEL_EXPORTER_OTLP_HEADERS="x-splyntra-api-key=${SPLYNTRA_API_KEY}"
export OTEL_SERVICE_NAME="my-agent"

Splyntra understands the OpenTelemetry GenAI semantic conventions, so LLM spans, token counts, and model names are picked up automatically. See Integrations for framework-specific notes.

3. See your first trace and risk score​

Run your agent once, then open the dashboard:

  • Traces — pick your latest run to see the full span tree with latency, tokens, and cost.
  • Risk — each run gets a risk score; drill into it to see which span triggered a secret, PII, or prompt-injection finding.
  • Cost — the run's spend is attributed per model and rolled up into the project.
{
  "run_id": "run_5f3a…",
  "project": "support-agent",
  "duration_ms": 4210,
  "cost_usd": 0.0132,
  "risk_score": 12,
  "findings": []
}

Next steps​