Bindu Agent Connectivity Layer

What does this agent do, and when should you use it?

Bindu is a developer framework for exposing an existing agent handler to an A2A network, rather than a replacement model or agent framework. Its Python, TypeScript, and Kotlin SDKs wrap a handler with bindufy() and expose A2A JSON-RPC, with the README’s examples using port 3773. The layer combines DID identity, Ed25519 message signatures, OAuth2 via Hydra, mTLS transport, and optional x402 USDC payments ahead of the handler. Submitted work can use message/send and be polled through tasks/get; the repository also lists Postgres storage, Redis scheduling, push notifications, private skills, and a gateway for orchestration. It fits teams that want to keep Agno, LangChain, CrewAI, or custom application logic while adding agent-to-agent connectivity and identity controls.

An application calls bindufy(config, handler) to register a handler that accepts messages and returns a response. The running service accepts A2A JSON-RPC operations including message/send, tasks/get, and message/stream; requests can carry X-DID-Signature, while the described security stack also uses mTLS certificates and OAuth2 bearer-token introspection through Ory Hydra. Bindu returns handler output as an A2A task response and lists Postgres for tasks and messages plus Redis-backed retries, timeouts, and recurring tasks. Setting deployment.url to a local address with expose: true opens an FRP public tunnel, while enabling x402 can require USDC on an EVM chain before the handler runs.

1. An Agno-based research-assistant developer who wants to expose an existing handler to other agents through A2A message/send.
2. A team operating a multi-agent workflow that needs the gateway to plan across an A2A fleet and stream the result back.
3. A provider of premium analysis or advisory work that wants to require USDC through x402 before business logic executes.
4. An enterprise integration team that needs DID identity, Ed25519 request signing, OAuth2 scopes, and mTLS around an agent endpoint.
5. A laptop-based developer who cannot configure port forwarding and wants expose: true to make a local deployment internet reachable.

What are this agent's strengths and limitations?

How do you install or deploy this agent?

Requires Python 3.12+ and uv. Install the SDK:

uv add bindu

For repository development:

git clone https://github.com/getbindu/Bindu.git
cd Bindu
uv sync --dev

The README examples also require at least one model-provider credential: OPENROUTER_API_KEY, OPENAI_API_KEY, or MINIMAX_API_KEY.

How do you use this agent?

The core Python integration is:

from bindu.penguin.bindufy import bindufy

config = {

"author": "[email protected]",
"name": "research_agent",
"description": "Research assistant with web search.",
"deployment": {"url": "http://localhost:3773", "expose": True},
"skills": ["skills/question-answering"],
}

def handler(messages: list[dict[str, str]]):

return agent.run(input=messages)

bindufy(config, handler)

The README’s Agno example connects that handler to OpenAIChat(id="gpt-4o"). Once live, POST a JSON-RPC message/send request to http://localhost:3773/ and poll tasks/get with the same taskId until its state is completed. The supplied material does not specify the exact command to launch the saved script or an installation command for the Agno dependency.

FAQ