voyager-help

name	voyager-help
description	Get help with Voyager SDK for Axelera AI hardware. Use when the user asks questions about Voyager SDK, Axelera hardware, pipelines, InferenceStream API, or model zoo. Explanation-only answers; no code changes or pipeline runs.
argument-hint	<question about Voyager SDK>
allowed-tools	Read, Bash, Glob, Grep, mcp__voyager__*

Voyager SDK Help

Get help with Voyager SDK for Axelera AI hardware

Important Context

Answer with explanations only; do not change code or run pipelines.

Instructions

Answer questions about Voyager SDK: $ARGUMENTS

Step 1: Data Source & Environment Selection

{{INCLUDE common/voyager-sdk-setup.md}}

Step 2: Answer the Question

After gathering information from the selected source:

1. Synthesize Response

   • Always frame answers in context of Axelera AI hardware
   • Provide Voyager SDK-specific code examples only after checking SDK/RAG evidence
   • Reference specific SDK components (InferenceStream, deploy.py, inference.py)
   • Distinguish documented, observed locally, and not validated here
   • Include a concise source note naming the SDK file, command output, or RAG result used

2. Include Practical Examples

   • Show actual command-line usage
   • Include YAML pipeline snippets when relevant
   • Reference model zoo locations (ax_models/zoo/)

3. Hardware Context

   • Mention detected Axelera device when relevant
   • Include AIPU core configuration options
   • Reference hardware capabilities such as vaapi, opencl, or opengl only when .voyager-runtime.json or SDK documentation supports them for this host

Common Topics

Topic	RAG Query	Key Files
Pipeline YAML	"pipeline YAML structure"	ax_models/zoo/**/*.yaml
InferenceStream	"InferenceStream API"	axelera/app/stream.py
Deployment	"model deployment compile"	deploy.py, axelera/app/compile.py
Input Sources	"USB RTSP video sources"	axelera/app/config.py
Custom Operators	"custom operator pipeline"	axelera/app/operators/
Trackers	"ByteTrack OC-SORT"	trackers/

Example Usage

voyager-help: How do I set up multi-stream inference?
voyager-help: What is the difference between the gst and torch-aipu pipes?
voyager-help: How to create a custom post-processor?
voyager-help: Explain the pipeline YAML structure
voyager-help: How to optimize for Axelera PCIe with 4 cores?