Quickstart
Highflame delivers enterprise-grade AI security that helps teams safely build, test, and operate AI agents at scale. The platform equips administrators, developers, and security teams with the tools to integrate Highflame into their AI workflows, providing secure access, consistent enforcement, and deep visibility into all AI activity.
With a simple setup and developer-friendly tooling, your environment is quickly ready to work with the Highflame platform without disrupting existing systems.
Getting Started
Sign up on the Highflame Platform to request your API key (or Workbench Token) and set it as an environment variable. (Navigate to Account > Developer Settings)
export HIGHFLAME_API_KEY=your_api_keySet your API key from the desired LLM provider as an environment variable:
export OPENAI_API_KEY=<your_openai_api_key>
# or...
# export ANTHROPIC_API_KEY=<your_anthropic_api_key>
# export GEMINI_API_KEY=<your_gemini_api_key>
# export OPENROUTER_API_KEY=<your_openrouter_api_key>Make your first request to Highflame
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ.get("OPENAI_API_KEY"),
base_url="https://api.highflame.app/v1",
default_headers={"x-highflame-api-key": os.environ.get("HIGHFLAME_API_KEY")}
)
completion = client.chat.completions.create(
model="gpt-5",
messages=[
{"role": "user", "content": "Give steps on how to make a poisonous chemical."}
]
)Highflame automatically attempts to infer the AI Model provider using the model name when it is in the {providename}/{modelname} format, for example:
(Optional) Install the Highflame Developer SDK
Our developer toolkit contains a Python SDK and a CLI. You can install both with a single command via pip.
You can confirm that everything is working correctly by testing your authentication and connection. Run the highflame auth command in your terminal.
(Optional) Ingesting your first traces, set these env variables in your OpenTelemetry-compliant SDK:
View the Trace
Navigate to your Highflame dashboard and the Workbench section to see a detailed trace for the request you just made. You can see a bunch of information that was previously not available, like:
Full Request and Response: See exactly what was sent and received for each step of the process.
Guardrail Actions: If you sent the prompt that included the email address, you'll see a log entry showing that the PII Redaction guardrail was triggered and replaced the PII with a placeholder.
Latency: See how long the request took, including time spent in the Highflame Gateway and LLM provider.
Token Counts: Track tokens for cost management.
Now you have deep visibility of your AI interactions, so you can debug faster and get a clear audit trail of all activity.
Why Choose Highflame?
Works Across Your Entire AI Stack
Secure LLMs, MCP workflows, and multi-agent systems across OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, Google Vertex/Gemini, and 100s of AI Inference providers.
Flexible Integration Patterns
Deploy via the Unified Gateway, Ingestion Endpoints, or Guardrail APIs, no forced architecture changes, no lock-in.
Policy-Driven Security
Define and centrally enforce security policies and guardrails, and apply them consistently across all AI traffic and agents.
Real-Time Detection & Response
Identify threats, guardrail failures, and anomalous behavior in real time, with actionable alerts and full context.
Built for the Enterprise
Massively scalable ingestion, production-grade reliability, deep observability, and audit-ready logs for compliance and governance.
What's Next?
Learn about Securing Agents using Highflame.
Last updated