Agen & Multi-Agen

Plano: AI-Native Proxy Server and Data Plane for Agentic Apps

katanemo/plano

Plano is an AI-native proxy server and data plane built in Rust that centralizes LLM routing, agent orchestration, observability, and guardrails, allowing developers to focus on core agent logic rather than infrastructure plumbing.

★ 7KBintang
⑂ 484Fork
138Isu terbuka
RustBahasa
Apache-2.0Lisensi
Q@project.QualityScoreSkor editorial

Tangkapan layar proyek

Tangkapan layar Plano: AI-Native Proxy Server and Data Plane for Agentic Apps Tangkapan layar Plano: AI-Native Proxy Server and Data Plane for Agentic Apps Tangkapan layar Plano: AI-Native Proxy Server and Data Plane for Agentic Apps Tangkapan layar Plano: AI-Native Proxy Server and Data Plane for Agentic Apps Tangkapan layar Plano: AI-Native Proxy Server and Data Plane for Agentic Apps

Ringkasan

Plano decouples agentic application delivery concerns from framework-specific code by moving them into a unified, out-of-process data plane. It handles low-latency orchestration between agents, smart LLM routing, zero-code capture of agentic signals and OpenTelemetry traces, and moderation hooks. Built on Envoy, Plano allows developers to declare agent descriptions and routing preferences in YAML instead of writing custom intent classifiers and routing logic.

Fitur utama

  • Low-latency agent orchestration via YAML configuration
  • Smart LLM routing by model name, alias, or preferences
  • Zero-code capture of Agentic Signals and OpenTelemetry traces/metrics
  • Moderation and memory hooks via Filter Chains
  • Framework-agnostic design supporting any language or AI framework
  • Built on Envoy proxy

Persyaratan, instalasi, dan mulai cepat

Follow the installation or quick-start section in the repository README and run its dependency setup commands. Repository: https://github.com/katanemo/plano

Penggunaan

To start using Plano, define your agents, model providers, and listeners in a 'config.yaml' file. Implement your agents as HTTP servers with an OpenAI-compatible '/v1/chat/completions' endpoint, pointing their LLM clients to Plano's local gateway. Start the Plano server using the command 'planoai up config.yaml'. You can then send chat completion requests to Plano's listener port, and it will automatically route the request to the appropriate agents.

Kompatibilitas model dan kasus penggunaan

Plano supports routing to models such as openai/gpt-4o and anthropic/claude-3-5-sonnet. It utilizes purpose-built lightweight LLMs, including a 4B-parameter Plano-Orchestrator model, for semantic routing.

Catatan lisensi dan risiko

Plano is licensed under the Apache-2.0 license.

Editorial verification 2026-08-09: repository URL, owner, description, license and repository statistics were reviewed. License metadata: Apache-2.0. README was fetched for the channel draft; re-check repository dependencies, releases and model terms before production use.

Rilis dan pemeliharaan

The repository configuration examples reference version v0.3.0. The project is actively maintained and not archived.

Firecrawl

firecrawl/firecrawl

★ 161,1KTypeScript

LangChain

langchain-ai/langchain

★ 143,6KPython

RAGFlow

infiniflow/ragflow

★ 86,7KGo