# Doca Gpi

> Use this skill for hands-on DOCA GPI programming — wiring a GPU-Packet-Initiator context so a CUDA kernel drives RDMA queues directly from GPU memory without host CPU mediation. Covers picking GPI vs doca-gpunetio, the docagpi / domain / channel object model, the GPU-side handle handoff (docagpugpichannel), attaching GPU memory to a GPI domain, the domain and channel attribute objects, and debugging DOCAERROR from docagpi calls. Trigger even when the user does not explicitly mention "DOCA GPI" — implicit phrasings include "my CUDA kernel needs to post RDMA directly from GPU memory", "DOCAERROR from docagpigpuchannelget", "how do I hand a GPU handle to my CUDA kernel", "how many channels can a GPI domain hold", or "GPU kernel driving RDMA without the host CPU on the path". Refuse and route elsewhere for the doca-gpunetio Send/Receive surface, the doca-rdma queue lifecycle, DPA-side initiation (doca-rdmi), or the CUDA programming model — those belong to other skills.

## Facts
- Page: https://tashan.sh/capability/skill-nvidia-doca-gpi
- tashan id: skill:NVIDIA/doca-gpi
- Source: https://github.com/NVIDIA/skills
- Type: skill
- Category: other
- tashan score: not scored (catalogued only — too little public evidence)
- Adoption: 9.0
- Upkeep: 96.0
- Freshness: 92.0
- Evidence coverage: 84% of the inputs this score can use
- Health: active
- Instruction depth: not yet graded
- License: Apache-2.0
- Official: no

## Install

```sh
cp -r doca-gpi ~/.claude/skills/
```

## Security audit
Not scanned. We audit npm-published capabilities; this one has no npm package we can resolve, or has not reached the queue. This is not a clean bill of health.

---
Measured 2026-08-20 by tashan (https://tashan.sh) from public evidence. Scorer s5.