Integrations
DeepintShield integrations: connect the AI stack you already run
DeepintShield integrations connect one gateway to the model providers, agent frameworks, identity systems and monitoring tools you already use. Most need one change: point an existing client at a new base URL with a virtual key, or add the Python SDK. Every listing below links to its documentation page.
One control point, connected to everything
Traffic from your apps, agents and coding tools passes through one gateway. Policy, identity, logging and cost controls apply on the way through, at the stages shown in how the request path works.
Model providers
- OpenAI
- Anthropic
- Azure
- AWS Bedrock
- Vertex AI
- Google Gemini
- Mistral
- Cohere
- Groq
- DeepSeek
- xAI
- Ollama and vLLM
- Custom OpenAI-compatible endpoints
Agent frameworks and SDKs
- LangChain and LangGraph
- OpenAI Agents SDK
- LlamaIndex
- AutoGen / AG2
- PydanticAI
- AWS Strands
- Google ADK
- LiteLLM
- Temporal
Identity
- Microsoft Entra ID
- Okta
- Google Workspace
- Auth0
- Keycloak
- Any OIDC provider

One control point
Observability and export
- OpenTelemetry
- Prometheus
- Grafana Cloud
- Datadog
- New Relic
- Honeycomb
- Langfuse
- S3, GCS and Azure Blob
- Snowflake, Redshift and BigQuery
Safety providers
- AWS Bedrock Guardrails
- Azure AI Content Safety
- Google Model Armor
- Your own webhook
Coding agents
- Claude Code
- Codex CLI
- Gemini CLI
- Cursor
- Qwen Code
- Opencode
- Zed
- Roo Code
Product names belong to their owners. A listing means a documented integration path, not a partnership.
29 built-in model providers behind one API
Pick a provider per request with a provider/model prefix, such as anthropic/claude-sonnet-4-5. All providers return responses in the OpenAI-compatible format, and provider credentials stay on the gateway. The AI Gateway adds weighted load balancing and fallbacks across them.
Custom providers add OpenAI-compatible or self-hosted endpoints, and passthrough routes send provider-native OpenAI, Anthropic or Gemini requests unchanged while governance and logging still apply.
- Frontier labs, cloud AI platforms and inference hosts
- Self-hosted runtimes: Ollama, vLLM and SGLang
- Speech, image and video providers
What each group of DeepintShield integrations adds
Every integration feeds the same policy, identity and evidence layer, so a new provider or framework inherits your existing controls.
Model providers
29 built-in providers and custom endpoints share load balancing, fallbacks, budgets and request logs.
AI Gateway
Agent frameworks and SDKs
Native binders for LangGraph, OpenAI Agents SDK, LlamaIndex, AutoGen, PydanticAI, Strands and Google ADK; tool calls are governed on Business and above.
Agentic Security
Identity
SSO over OIDC with Entra ID, Okta, Google Workspace or Auth0, plus JIT provisioning and SCIM sync with Entra ID.
Governance
Observability and export
OpenTelemetry with GenAI conventions, Prometheus metrics, and scheduled log exports to cloud storage and data warehouses.
Observability & Audit
Safety providers
Run AWS Bedrock Guardrails, Azure AI Content Safety, Google Model Armor or a webhook in one policy, on Team and above.
Runtime Guardrails & PII
MCP servers and coding agents
Connect MCP servers over HTTP or SSE with OAuth 2.1, and point Claude Code, Codex CLI or Cursor at the gateway.
MCP Gateway
All integrations
Filter by group to find your provider, framework, identity system or monitoring tool. Each card links to its documentation page, and the docs keep a full list of providers and frameworks.
Azure OpenAI Service
OpenAI and Claude deployments on Azure via API key, Entra ID or Managed Identity.
AWS Bedrock
Claude, Nova, Llama, Mistral, Cohere and Titan using your AWS credentials or IAM role.
Amazon Bedrock Mantle
Bedrock's OpenAI-compatible endpoint: chat, Responses, embeddings and files.
Vertex AI
Gemini and Claude on Google Cloud with service-account auth, Imagen and Veo.
Google Gemini
Gemini chat, embeddings, Imagen images, Veo video, speech, files and batch.
OpenAI SDK
Point the OpenAI client at DeepintShield to reach all 29 provider identities.
AWS Bedrock SDK (boto3)
Point boto3 bedrock-runtime at DeepintShield for Converse and Invoke calls.
Passthrough
Send provider-native OpenAI, Anthropic, Gemini or Vertex requests unchanged.
LangChain and LangGraph
Native ChatOpenAI models and embeddings routed through the gateway.
AutoGen / AG2
Native binder supplies AutoGen model clients routed through the gateway.
Temporal
Call the gateway from Temporal activities; Temporal keeps workflow execution.
Log exports
Scheduled log exports to S3, GCS, Azure Blob, Snowflake, Redshift or BigQuery.
Durable webhooks
Signed, at-least-once delivery of control-plane events to your endpoints.
AWS Bedrock Guardrails
Run AWS Bedrock Guardrails as a safety provider inside DeepintShield policies.
Azure AI Content Safety
Run Azure AI Content Safety as a safety provider inside DeepintShield policies.
Google Model Armor
Run Google Model Armor as a safety provider inside DeepintShield policies.
Custom webhook safety provider
Bring your own detector to DeepintShield policies through a webhook.
Most integrations are one line
Keep your client and change where it points. Callers authenticate with a DeepintShield virtual key, so provider keys never sit in application code. Compatibility is route-specific, so test the exact endpoints and SDK version your application uses.
- MCP clients: the gateway’s /mcp endpoint with a virtual key
- OpenTelemetry: OTLP over HTTP or gRPC
- Python SDK 3.2.3 for Python 3.10 or newer
import openai
client = openai.OpenAI(
base_url="https://app.deepintshield.com/openai",
api_key="sk-ds-your-virtual-key"
)import anthropic
client = anthropic.Anthropic(
base_url="https://app.deepintshield.com/anthropic",
api_key="sk-ds-your-virtual-key",
default_headers={"x-deepintshield-vk": "sk-ds-your-virtual-key"},
)from deepintshield import DeepintShield
with DeepintShield.from_env() as shield:
llm = shield.bind("langchain").model("anthropic/claude-sonnet-4-5")
response = llm.invoke("Explain retrieval-augmented generation in one sentence.")
print(response.content)import os
from litellm import completion
response = completion(
model="openai/anthropic/claude-sonnet-4-5",
custom_llm_provider="openai",
api_base="https://app.deepintshield.com/litellm",
api_key=os.environ["DEEPINTSHIELD_VIRTUAL_KEY"],
messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)Need an integration that is not listed? Tell us what you run and an engineer will check it with you.
Connect your stack with one base URL
Start free with 50,000 requests a month, or book a demo to see your own providers and frameworks behind one gateway.
