Frameworks

TapPass works with any framework that speaks OpenAI or Anthropic protocol. Most integrations are a single environment variable. The rest are a few lines of code.

At a glance

Framework	Language	Method	Setup time
OpenAI SDK	Python, Node, Go, Ruby, …	Env var	30s
Anthropic SDK	Python, Node	Env var	30s
LangChain	Python	guard_agent()	2 min
CrewAI	Python	guard_crew()	2 min
LlamaIndex	Python	Env var	30s
Pydantic AI	Python	Env var	30s
Google ADK	Python	guard_adk_agent()	2 min
FastAPI	Python	Middleware	3 min
Vercel AI SDK	TypeScript	Env var	30s
MCP	Any	MCP server	5 min

Zero-code: two env vars

Any OpenAI- or Anthropic-compatible tool works without a code change:

# OpenAI-compatible (Cursor, Copilot, LangChain, LlamaIndex, CrewAI, Pydantic AI, …)
export OPENAI_BASE_URL=https://tappass.example.com/v1
export OPENAI_API_KEY=tp_...

# Anthropic-compatible (Claude Code, Cline, Anthropic SDK)
export ANTHROPIC_BASE_URL=https://tappass.example.com
export ANTHROPIC_API_KEY=tp_...

Python SDK (when you need more)

import tappass
from tappass import Agent, AsyncAgent

agent = Agent("https://tappass.example.com", "tp_...")

# Governed chat
response = agent.chat("What are the GDPR requirements?")

# Streaming
for chunk in agent.stream("Write a compliance report"):
    print(chunk, end="", flush=True)

# Wrap existing tools (LangChain / CrewAI / LlamaIndex / Pydantic AI tools)
governed_tools = tappass.govern(
    [search, send_email, query_db],
    url="https://tappass.example.com",
    api_key="tp_...",
    mode="enforce",
)

# Async
async with AsyncAgent("https://tappass.example.com", "tp_...") as agent:
    response = await agent.chat("Analyze the data")
    async for chunk in agent.stream("Summarize"):
        print(chunk, end="", flush=True)

See the SDK Reference for the full surface.

Examples repository

Working end-to-end examples for every framework live in tappass/tappass-examples.