doca-spcx-cc
Use this skill when the user is invoking `doca_spcx_cc` (the host-side CLI under /opt/mellanox/doca/tools/) to load, parameterize, start, observe, or stop a Programmable Congestion Control (SPCX) algorithm on a BlueField with a DPA processor against a live RDMA / RoCE fabric, or picking SPCX vs the established `doca-pcc` surface. Trigger even when the user does not say "DOCA SPCX" or "doca_spcx_cc" — typical implicit phrasings include "I want to write a custom RTT-based CC algorithm for my RoCE fabric", "my SPCX session loaded but throughput / latency didn't change", "doca_pcc status shows A
npx skills add NVIDIA/skills --skill doca-spcx-cc --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
This skill provides a host-side tool pathway to work with SPCX-class programmable congestion control by orchestrating the CLI tool that loads, parameterizes, starts, observes, and stops SPCX algorithms on a BlueField with a DPA processor, in comparison with the PCC surface where available.
How it works
It describes usage of the host CLI doca_spcx_cc (the host-side CLI under /opt/mellanox/doca/tools/) to manage SPCX algorithms on the DPA-enabled BlueField. It emphasizes loading an authored SPCX algorithm (or a reference) onto the BlueField via the SPCX surface, parameterizing it, starting it, observing its effect on the live RDMA/RoCE link, and evaluating performance against the factory PCC path. It also provides guidance on deciding between SPCX and doca-pcc, and on safety, observability, and rollback plans. The skill acts as guidance for workflows such as load, configure, run, test, debug, and use, and directs users to CAPABILITIES.md and TASKS.md for concrete steps and limitations.
When to use it
Use when you need to invoke doca_spcx_cc on a DOCA-installed host with a BlueField DPA processor and a non-prod RDMA/RoCE fabric, to load/parameterize/start/observe an SPCX algorithm, or to compare SPCX vs PCC paths and to form a safe-to-roll-forward decision with evidence and rollback plans.
What it can touch
The skill references the host-side CLI doca_spcx_cc, the DOCA SPCX path, and its interaction with the doca-pcc library. It notes checks such as the preconditions for hardware/firmware availability, version compatibility, and the live-link performance surface, but does not specify exact commands beyond mentioning the CLI binary and its usage surface.
Caveats
It notes that substantive material lives in CAPABILITIES.md and TASKS.md, and that the skill does not ship a specific CC algorithm, pre-baked outputs, or scripting wrappers. It requires a DOCA-enabled host, compatible BlueField with a DPA processor and a firmware custom-PCC slot, and a non-prod fabric with controllable contention. It also emphasizes safety policies and rollback procedures rather than autonomous recommendations.
# DOCA SPCX Congestion-Control Tool **Where to start:** This is a tool skill for invoking `doca_spcx_cc` — the documented host-side CLI that exercises an SPCX-class Programmable Congestion Control algorithm on a live RDMA / RoCE fabric driven by a BlueField with a DPA processor. Open [`TASKS.md`](TASKS.md) and start at [`## configure`](TASKS.md#configure) for the SPCX-vs-PCC-vs-factory-firmware de
What does the doca-spcx-cc skill do?
Use this skill when the user is invoking `doca_spcx_cc` (the host-side CLI under /opt/mellanox/doca/tools/) to load, parameterize, start, observe, or stop a Programmable Congestion Control (SPCX) algorithm on a BlueField with a DPA processor against a live RDMA / RoCE fabric, or picking SPCX vs the established `doca-pcc` surface. Trigger even when the user does not say "DOCA SPCX" or "doca_spcx_cc" — typical implicit phrasings include "I want to write a custom RTT-based CC algorithm for my RoCE fabric", "my SPCX session loaded but throughput / latency didn't change", "doca_pcc status shows A
How do I install it?
Run `npx skills add NVIDIA/skills --skill doca-spcx-cc --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From NVIDIA/skills, a repository with 2,789 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.