dora-router

name	dora-router
description	CRITICAL: Use for ALL dora-rs questions including dataflow config, node development, and robotic applications. Triggers on: dora, dora-rs, dataflow, DoraNode, EventStream, dora run, dora build, dataflow.yml, nodes:, inputs:, outputs:, dora/timer, shared memory, arrow, 机器人, robotics, 数据流, 节点, node api, operator, python node, rust node, camera, yolo, detection, vision, audio, speech, arm control, lerobot
globs	["**/dataflow.yml","**/dataflow.yaml","**/*.py","**/*.rs"]
source	https://dora-rs.ai

Dora Question Router

Version: 1.0.0 | Last Updated: 2026-01-20

Intelligent routing for dora-rs dataflow development questions

Overview

dora-rs is a Dataflow-Oriented Robotic Architecture framework. This router helps navigate to the correct skill based on the question type.

Question Type Routing

Core Development

Question Pattern	Keywords	Route To
Dataflow YAML configuration	nodes, inputs, outputs, timer, yaml	dataflow-config
Rust node development	DoraNode, EventStream, send_output, init_from_env	node-api-rust
Python node development	dora.Node, event, python node	node-api-python
Operator development	DoraOperator, on_event, register_operator	operator-api
CLI commands	dora run, dora build, dora start, dora stop	cli-commands
Testing nodes	integration testing, test inputs, JSONL	integration-testing

Domain Applications

Domain Keywords	Route To
YOLO, detection, segmentation, VLM, camera, image	domain-vision
speech, TTS, STT, voice, audio, Whisper, Kokoro	domain-audio
arm control, chassis, actuator, robot, servo	domain-robot
recording, replay, lerobot, data collection	data-pipeline
dora-yolo, dora-rerun, opencv-video-capture	hub-nodes

Hub Node Skills (Detailed)

Question About	Keywords	Route To
Camera/video capture	opencv-video-capture, realsense, orbbeck, webcam	hub-camera
Audio processing	microphone, vad, whisper, pyaudio, kokoro-tts	hub-audio
Object detection/tracking	yolo, sam2, cotracker, bounding box, mask	hub-detection
Language models	qwen, vlm, llm, internvl, text generation	hub-llm
Robot control	piper, reachy, ugv, kit-car, rdt-1b, arm	hub-robot
Visualization	rerun, opencv-plot, display, primitives	hub-visualization
Data recording	llama-factory-recorder, lerobot-dashboard	hub-recording
Translation	opus, argotranslate, translate	hub-translation

Routing Examples

User: "How do I create a dataflow YAML?"
Route: dataflow-config

User: "How to send output from a Rust node?"
Route: node-api-rust

User: "I want to add YOLO detection to my pipeline"
Route: domain-vision + hub-nodes

User: "How to control a robot arm with dora?"
Route: domain-robot

User: "How to record data for training?"
Route: data-pipeline

Architecture Overview

┌─────────────────────────────────────────────────┐
│                  Dataflow YAML                   │
│  (nodes, inputs, outputs, connections)          │
└─────────────────────────────────────────────────┘
                        │
        ┌───────────────┼───────────────┐
        ▼               ▼               ▼
┌───────────────┐ ┌───────────────┐ ┌───────────────┐
│  Python Node  │ │   Rust Node   │ │   Operator    │
│  (dora.Node)  │ │  (DoraNode)   │ │(DoraOperator) │
└───────────────┘ └───────────────┘ └───────────────┘
        │               │               │
        └───────────────┼───────────────┘
                        ▼
┌─────────────────────────────────────────────────┐
│           Shared Memory / Arrow Data            │
│         (Zero-copy data transfer)               │
└─────────────────────────────────────────────────┘

Quick Reference

CLI Commands

Command	Description
dora new	Create new dataflow/node/operator
dora build	Build nodes in a dataflow
dora run	Run dataflow (standalone)
dora up	Start coordinator and daemon
dora start	Start dataflow on daemon
dora stop	Stop running dataflow
dora list / dora ps	List running dataflows
dora logs	View node logs
dora destroy	Stop coordinator and daemon

Timer Input Format

inputs:
  tick: dora/timer/millis/100  # Every 100ms
  tick: dora/timer/secs/5      # Every 5 seconds

Input Connection Format

inputs:
  input_name: source_node/output_name

Skill Files

Core Skills

skills/dataflow-config/SKILL.md     # YAML configuration
skills/node-api-rust/SKILL.md       # Rust DoraNode API
skills/node-api-python/SKILL.md     # Python dora.Node API
skills/operator-api/SKILL.md        # Operator development
skills/cli-commands/SKILL.md        # CLI usage
skills/integration-testing/SKILL.md # Testing nodes

Domain Skills

skills/domain-vision/SKILL.md       # ML/Vision pipelines
skills/domain-audio/SKILL.md        # Audio processing
skills/domain-robot/SKILL.md        # Robot control
skills/data-pipeline/SKILL.md       # Data collection
skills/hub-nodes/SKILL.md           # Pre-built nodes overview

Hub Node Skills (Detailed)

skills/hub-camera/SKILL.md          # Camera capture nodes
skills/hub-audio/SKILL.md           # Audio processing nodes
skills/hub-detection/SKILL.md       # Object detection/tracking
skills/hub-llm/SKILL.md             # Language models
skills/hub-robot/SKILL.md           # Robot control nodes
skills/hub-visualization/SKILL.md   # Rerun and OpenCV plot
skills/hub-recording/SKILL.md       # Data recording nodes
skills/hub-translation/SKILL.md     # Translation nodes

Instructions for Claude

When handling dora-rs questions:

1. Identify Question Type - Match keywords to routing table
2. Load Appropriate Skill - Read the skill file for context
3. Provide Examples - Include code examples from skills
4. Reference Hub Nodes - Suggest pre-built nodes when applicable

Default Project Settings

When creating dora projects:

• Use latest dora-rs version (0.4.0)
• Use edition = "2024" for Rust nodes
• Use Apache Arrow for data format
• Configure appropriate timer frequencies

Priority Order

1. Check if question is about dataflow configuration
2. Check if question is about specific API (Rust/Python/Operator)
3. Check if question is domain-specific (vision/audio/robot)
4. Suggest relevant hub nodes for common tasks