DOCA STA (Storage Target Acceleration)
Examples
- "Offload the nvmf target onto the DPU." — this is in
scope. Stand up
doca_sta + subsystem/namespaces; do not
rewrite the initiator stack here.
- "Identify Controller times out over RoCE." — check
doca_sta_cap_is_supported and queue sizing before
changing backend disks.
Instructions
Follow Loading order below. Open the tasks companion for configure /
target / debug. Open the capabilities companion for queue limits
and NVMe-oF caps. Do not open every companion file.
Where to start: This skill assumes DOCA is already installed and
the user is doing hands-on NVMe-over-Fabrics storage-target work on
a BlueField-class device with DOCA. Open the tasks companion if
the user wants to do something (configure / modify / build / run
/ test / debug); open the capabilities companion when the
question is what can DOCA STA express on this version. If the
user has not installed DOCA yet, route to
doca-setup/SKILL.md first. If the user is
asking "is this an NVMe-oF initiator/host transport?", the
answer is no — doca-sta accelerates the target side: it presents
NVMe-oF doca_sta_subsystem targets backed by local NVMe-PCI
disks; the model lives in
the capabilities companion.
Example questions this skill answers well
The CLASSES of DOCA STA questions this skill is built to answer,
each with one worked example. The agent should treat the class
as the load-bearing piece — the worked example is a single
instance.
- "How do I bring up an NVMe-oF target that uses the BlueField
to accelerate the storage data path?" — worked example: "define
a
doca_sta_subsystem (NQN) with one namespace backed by a local
NVMe-PCI disk (doca_sta_be) and accept NVMe-over-RDMA
connections from a remote initiator". Answered by the
target-model-and-lifecycle
workflow in the tasks companion +
the capabilities companion
target-object table.
- "Can this BlueField accelerate an NVMe-oF target at all?" —
worked example: "my data center is RoCE end-to-end; does this
device support DOCA STA target acceleration?". STA transport is
RDMA-only (there is no NVMe-over-TCP path). Answered by the
capability-query rule (
doca_sta_cap_is_supported against a
doca_devinfo) in
the capabilities companion
- the discovery step in
the tasks companion.
- "How deep can I size my I/O queues, and how many I/O queues
per connection?" — worked example: "I want 16 I/O queues at
depth 1024 each — does this device support that?". Answered by
the queue-sizing capability surface in
the capabilities companion
- the queue-sizing step in
the tasks companion which gates on
the matching
doca_sta_get_max_* query (e.g.
doca_sta_get_max_qps, doca_sta_get_max_io_queue_size).
- "Which other DOCA libraries do I need alongside doca-sta?" —
worked example: "do I need doca-rdma directly, or does doca-sta
hide it from me?". Answered by the substrate-library rule in
the capabilities companion
- the env-prep checklist in
the tasks companion step 1, which
routes the steering side to
doca-flow/SKILL.md and the RDMA substrate
to doca-rdma/SKILL.md.
- "Is this STA capability available on my installed DOCA?" —
worked example: "is STA target acceleration supported on this
BlueField + DOCA version?". Answered by the version-and-device
overlay in
the capabilities companion,
which cross-links the canonical detection chain in
doca-version/SKILL.md and adds the
STA-specific cap-query rule (pkg-config --modversion doca-sta
is the build-time anchor; the runtime doca_sta_cap_is_supported
query is the truth).
- "What does this
DOCA_ERROR_* from a STA call mean and which
layer caused it?" — worked example: "DOCA_ERROR_NOT_PERMITTED
when opening the doca_dev".
Answered by the STA overlay on the cross-library taxonomy in
the capabilities companion
- the layered ladder in
the tasks companion that escalates to
doca-debug/SKILL.md.
A Connect handshake that never completes is usually not an STA
bug. It is the single most common first-app failure, and the
answer is to walk the three preconditions in
the capabilities companion
before any code change: the RDMA substrate resolves on the chosen
device, the doca_dev was opened with sufficient permission, and —
most often the culprit — a DOCA Flow rule actually steers the
matching NVMe-oF 5-tuples to this instance's queues. The steering
side belongs to doca-flow/SKILL.md; route it
rather than inventing a doca_sta_* steering call, and do not
start by auditing STA lifecycle or target-side code. The same
applies to an Identify Controller that times out over RoCE.
DOCA_ERROR_IO_FAILED on a per-I/O completion is a transport
error, not a target-side STA bug — a link drop, RDMA peer
disconnect, firmware fault, or peer controller reset. Being able to
ping the target does not exclude any of these. Do not retry blindly
and do not change STA code first: capture dmesg and route to
doca-rdma/SKILL.md, per
the tasks companion.
Audience
This skill serves external developers building NVMe-over-Fabrics
storage targets that consume DOCA STA on BlueField — i.e., users
whose code calls doca_sta_* (directly in C/C++, or through
FFI/bindings from another language) to accelerate the target-side
data path of an NVMe-oF target on the BlueField hardware: presenting
doca_sta_subsystem targets (NQN + namespaces) backed by local
NVMe-PCI disks (doca_sta_be) to remote initiators over RDMA. The
skill is not for NVIDIA developers contributing to DOCA STA
itself, and it is not for initiator/host-side NVMe stacks.
Language scope. DOCA STA ships as a C library with
pkg-config module name doca-sta. DOCA STA ships no public
samples — the public packages install no
samples/doca_sta/ directory — so the worked examples in
the tasks companion build against the public headers directly rather than
modify a shipped sample. C and C++ consumers are the canonical
case. Other-language
consumers (Rust, Go, Python, …) consume the same *.so through
FFI or language-specific bindings; the skill's contribution in
that case is to keep the target-model, lifecycle,
capability-discovery, queue-pair shape, substrate-dependency,
and error-taxonomy guidance language-neutral, and to route the
agent to the public C ABI as the authoritative surface that any
wrapper will eventually call.
When to load this skill
Load this skill when the user is doing hands-on DOCA STA work,
in any language. Concretely:
- Initializing a
doca_sta instance on a doca_dev opened
against a BlueField PF / SF and configuring the NVMe-oF
target subsystems before doca_ctx_start().
- Defining target resources —
doca_sta_subsystem (NQN +
namespaces) and doca_sta_be backend controllers (local
NVMe-PCI disks) — and accepting NVMe-oF connections (admin
queue plus N I/O queues per connection) on the target side as
remote initiators connect over RDMA CM.
- Reading or setting STA properties via the
doca_sta_set_*
family, checking device support via doca_sta_cap_is_supported,
and querying sizing limits via the doca_sta_get_max_* family
(max I/O queue depth, max number of queue pairs, max I/O size,
max subsystems, max namespaces per subsystem, max backends).
- Wiring the NVMe-over-RDMA transport — STA's only transport;
it lands on the
doca-rdma substrate and uses RDMA CM for
connection establishment — for the target's I/O queues.
- Wiring DOCA Flow rules so that NVMe-oF traffic actually
reaches the STA-managed queues — the steering boundary is
doca-flow, not doca-sta.
- Debugging a
DOCA_ERROR_* returned from a STA call (lifecycle
vs. capability vs. transport-layer I/O failure vs.
driver-below) and the per-queue events on the DOCA Core
progress engine.
- Designing or extending non-C bindings (Rust, Go, Python, …)
that wrap the DOCA STA C ABI — for the lifecycle, queue-pair,
cap-query, and substrate-dependency rules the wrapper must
honor.
Do not load this skill for general DOCA orientation, install
of DOCA itself, raw RDMA data movement (use
doca-rdma/SKILL.md), raw packet I/O on
Ethernet queues (use doca-eth/SKILL.md),
flow-rule programming (use doca-flow/SKILL.md),
or initiator/host-side NVMe stack development
(SPDK or kernel-nvme own that, not this skill). For DOCA
documentation orientation, use
doca-public-knowledge-map/SKILL.md.
What this skill provides
This is a thin loader. The body keeps only the orientation
needed to pick the right next file. The substantive STA-specific
material lives in two companion files:
- the capabilities companion — what DOCA STA can express on this
version: the target object model (
doca_sta_subsystem /
namespaces / doca_sta_be backend NVMe-PCI disks),
the NVMe queue-pair shape (admin queue + I/O queues over RDMA),
the RDMA-only transport,
the capability-query surface (doca_sta_cap_is_supported plus
the doca_sta_get_max_* sizing queries), the STA
error taxonomy (mapped onto the cross-library DOCA_ERROR_*
set), the observability surface (per-queue progress engine
events, capability snapshots), and the safety policy that
gates substrate-library, permission, and steering
preconditions.
- the tasks companion — step-by-step workflows for the six in-scope
STA verbs:
configure, modify, build, run, test,
debug. Plus a Deferred task verbs block that points
out-of-scope questions at the right next skill, and a
Command appendix of the recurring commands the agent
reaches for.
The skill assumes a BlueField (with DOCA installed at the
standard location) plus a remote NVMe-oF initiator reachable on
the fabric to connect into the accelerated target, and one or
more local NVMe-PCI disks to back the target's namespaces. It
does not cover installing
DOCA — that path goes through
doca-setup/SKILL.md. It does not cover
initiator/host-side NVMe stacks (SPDK bdev_nvme, kernel nvme
host) or NVMe protocol semantics above the accelerated target
data path — those are out of scope.
What this skill deliberately does not ship
This skill is agent guidance, not a samples or templates
bundle. To keep the boundary clean, it deliberately does not
contain — and pull requests should not add:
- Pre-written DOCA STA application source code, in any
language. DOCA STA ships no public samples — there is no
/opt/mellanox/doca/samples/doca_sta/ directory. The authoritative surface is the public headers under
$(pkg-config --variable=includedir doca-common) plus the public
DOCA STA guide; the agent builds against those directly rather
than modifying a shipped sample, per the
the tasks companion workflow.
- Initiator/host-side NVMe stack glue. SPDK
bdev_nvme,
the kernel nvme host, and any initiator-side NVMe stack are
upstream projects out of scope for this skill — DOCA STA is
target-side acceleration, not an initiator transport provider.
- Standalone build manifests (
meson.build, CMakeLists.txt,
Cargo.toml, …) parked inside the skill. The agent
constructs the build manifest in the user's project
directory against the user's installed DOCA, where
pkg-config --modversion doca-sta is the source of truth.
- A
samples/, bindings/, or reference/ subtree of any
kind. A mock or incomplete artifact in this skill's tree,
even one labeled "reference", is misleading: users will
read it as buildable.
Loading order
- Read this skill first to confirm the user's question
is in scope.
- For the STA capability matrix, the target object model,
queue-pair shape, RDMA-only transport, capability-query
rules, error taxonomy, observability, and safety policy,
see the capabilities companion.
- For step-by-step workflows — configure, modify, build,
run, test, debug — see the tasks companion.
Both companion files cross-link to each other,
doca-version/SKILL.md for the canonical
version-handling rules,
doca-rdma/SKILL.md for the RDMA substrate
that NVMe-over-RDMA transport lands on,
doca-flow/SKILL.md for the steering rules
that direct NVMe traffic to STA-managed queues, and
doca-public-knowledge-map/SKILL.md
whenever the right answer is "look it up in the public docs or
the installed package layout" rather than "STA-specific
guidance".
Related skills
doca-public-knowledge-map/SKILL.md —
the routing table for every public DOCA documentation source
and the on-disk layout of an installed DOCA package. The
STA URL slug is DOCA-STA.
doca-setup/SKILL.md — env preparation,
install verification, BlueField mode checks, and the
permission / group-membership requirements for opening a
doca_dev. This skill assumes its preconditions are
satisfied.
doca-version/SKILL.md — canonical
DOCA version-handling rules. This skill's ## Version compatibility cross-links the four-way match rule and adds
only the STA-specific overlay (STA target-acceleration
availability windows, NVMe-oF feature-set device-conditional
support).
doca-structured-tools-contract/SKILL.md —
the bundle's structured-tools precedence rule (detect /
prefer / fall back / report). The Command appendix in
the tasks companion honors this contract.
doca-programming-guide/SKILL.md —
general DOCA programming patterns shared by every library:
the canonical pkg-config + meson build pattern, the
universal modify-a-shipped-sample first-app workflow, the
universal lifecycle, the cross-library DOCA_ERROR_*
taxonomy, and the program-side debug order. This skill
layers STA specifics on top.
doca-rdma/SKILL.md — the RDMA substrate
that NVMe-over-RDMA transport lands on. STA hides most of
the RDMA queue-pair details from the consumer, but the user
still needs doca-rdma linked in and the device's RDMA
capabilities discoverable for the NVMe-over-RDMA path to
work.
doca-eth/SKILL.md — the queue-pair
shape that STA's per-connection queue model echoes. Reach
here if the user is asking general questions about how
DOCA exposes queue-pairs that don't have an STA-specific
answer.
doca-flow/SKILL.md — the steering
surface that decides which NVMe-oF packets land on which
STA-managed queue. DOCA STA does not program steering
itself; an NVMe-oF target whose connections never come up
is often a missing or wrong Flow rule, not a STA bug.
doca-debug/SKILL.md — the
cross-cutting debug ladder (install / version / build /
link / runtime / program / driver). STA-specific debug
(transport-type mismatches, queue-depth oversize,
IO-failed transport errors) overlays on top of that
ladder.
1---2license: Apache-2.03name: doca-sta4description: >5 Use this skill when the user builds an NVMe-over-Fabrics6 storage target on BlueField with DOCA STA — instantiating7 doca_sta, declaring doca_sta_subsystem NQNs and namespaces over8 local NVMe disks via doca_sta_be, validating platform support9 with doca_sta_cap_is_supported, sizing I/O queue depth, or10 decoding DOCA_ERROR_* statuses such as IO_FAILED on NVMe reads.11 Also load it for implied asks like "Identify Controller times out12 over RoCE", "offload the nvmf target", "NVMe-oF Connect never13 completes", or "16 I/O queues at depth 1024 — supported?". Refuse14 and route to the proper owner for raw RDMA movement, packet I/O,15 steering rules, initiator-side NVMe, or SDK install.16metadata:17 source: https://skillmd.com/skills/nvidia/doca-sta18 author: NVIDIA <doca-feedback@nvidia.com>19 kind: library20 doca_version: "3.5"21 tags:22 - doca23 - sta24 - nvme25compatibility: >26 Requires DOCA SDK installed at /opt/mellanox/doca on Linux (Ubuntu27 22.04/24.04 or RHEL/SLES) with a BlueField DPU or ConnectX NIC28 attached. Reads the user's local install via `pkg-config doca-sta`29 (and `pkg-config doca-rdma` for the NVMe-over-RDMA transport) and30 inspects /opt/mellanox/doca/{lib,include,samples,applications}. The31 validation steps require a live DOCA install.32---3334# DOCA STA (Storage Target Acceleration)3536## Examples3738- **"Offload the nvmf target onto the DPU."** — this is in39 scope. Stand up `doca_sta` + subsystem/namespaces; do not40 rewrite the initiator stack here.41- **"Identify Controller times out over RoCE."** — check42 `doca_sta_cap_is_supported` and queue sizing before43 changing backend disks.4445## Instructions4647Follow Loading order below. Open the tasks companion for configure /48target / debug. Open the capabilities companion for queue limits49and NVMe-oF caps. Do not open every companion file.5051**Where to start:** This skill assumes DOCA is already installed and52the user is doing **hands-on NVMe-over-Fabrics storage-target work** on53a BlueField-class device with DOCA. Open the tasks companion if54the user wants to *do* something (configure / modify / build / run55/ test / debug); open the capabilities companion when the56question is *what can DOCA STA express* on this version. If the57user has not installed DOCA yet, route to58`doca-setup/SKILL.md` first. If the user is59asking *"is this an NVMe-oF initiator/host transport?"*, the60answer is no — doca-sta accelerates the **target** side: it presents61NVMe-oF `doca_sta_subsystem` targets backed by local NVMe-PCI62disks; the model lives in63the capabilities companion.6465## Example questions this skill answers well6667The CLASSES of DOCA STA questions this skill is built to answer,68each with one worked example. The agent should treat the *class*69as the load-bearing piece — the worked example is a single70instance.7172- **"How do I bring up an NVMe-oF target that uses the BlueField73 to accelerate the storage data path?"** — worked example: *"define74 a `doca_sta_subsystem` (NQN) with one namespace backed by a local75 NVMe-PCI disk (`doca_sta_be`) and accept NVMe-over-RDMA76 connections from a remote initiator"*. Answered by the77 target-model-and-lifecycle78 workflow in the tasks companion +79 the capabilities companion80 target-object table.81- **"Can this BlueField accelerate an NVMe-oF target at all?"** —82 worked example: *"my data center is RoCE end-to-end; does this83 device support DOCA STA target acceleration?"*. STA transport is84 RDMA-only (there is no NVMe-over-TCP path). Answered by the85 capability-query rule (`doca_sta_cap_is_supported` against a86 `doca_devinfo`) in87 the capabilities companion88 + the discovery step in89 the tasks companion.90- **"How deep can I size my I/O queues, and how many I/O queues91 per connection?"** — worked example: *"I want 16 I/O queues at92 depth 1024 each — does this device support that?"*. Answered by93 the queue-sizing capability surface in94 the capabilities companion95 + the queue-sizing step in96 the tasks companion which gates on97 the matching `doca_sta_get_max_*` query (e.g.98 `doca_sta_get_max_qps`, `doca_sta_get_max_io_queue_size`).99- **"Which other DOCA libraries do I need alongside doca-sta?"** —100 worked example: *"do I need doca-rdma directly, or does doca-sta101 hide it from me?"*. Answered by the substrate-library rule in102 the capabilities companion103 + the env-prep checklist in104 the tasks companion step 1, which105 routes the steering side to106 `doca-flow/SKILL.md` and the RDMA substrate107 to `doca-rdma/SKILL.md`.108- **"Is this STA capability available on my installed DOCA?"** —109 worked example: *"is STA target acceleration supported on this110 BlueField + DOCA version?"*. Answered by the version-and-device111 overlay in112 the capabilities companion,113 which cross-links the canonical detection chain in114 `doca-version/SKILL.md` and adds the115 STA-specific cap-query rule (`pkg-config --modversion doca-sta`116 is the build-time anchor; the runtime `doca_sta_cap_is_supported`117 query is the truth).118- **"What does this `DOCA_ERROR_*` from a STA call mean and which119 layer caused it?"** — worked example: *"`DOCA_ERROR_NOT_PERMITTED`120 when opening the `doca_dev`"*.121 Answered by the STA overlay on the cross-library taxonomy in122 the capabilities companion123 + the layered ladder in124 the tasks companion that escalates to125 `doca-debug/SKILL.md`.126127> **A Connect handshake that never completes is usually not an STA128> bug.** It is the single most common first-app failure, and the129> answer is to walk the three preconditions in130> the capabilities companion131> *before* any code change: the RDMA substrate resolves on the chosen132> device, the `doca_dev` was opened with sufficient permission, and —133> most often the culprit — a DOCA Flow rule actually steers the134> matching NVMe-oF 5-tuples to this instance's queues. The steering135> side belongs to `doca-flow/SKILL.md`; route it136> rather than inventing a `doca_sta_*` steering call, and do not137> start by auditing STA lifecycle or target-side code. The same138> applies to an Identify Controller that times out over RoCE.139140> **`DOCA_ERROR_IO_FAILED` on a per-I/O completion is a transport141> error, not a target-side STA bug** — a link drop, RDMA peer142> disconnect, firmware fault, or peer controller reset. Being able to143> ping the target does not exclude any of these. Do not retry blindly144> and do not change STA code first: capture `dmesg` and route to145> `doca-rdma/SKILL.md`, per146> the tasks companion.147148## Audience149150This skill serves **external developers building NVMe-over-Fabrics151storage targets that consume DOCA STA on BlueField** — i.e., users152whose code calls `doca_sta_*` (directly in C/C++, or through153FFI/bindings from another language) to accelerate the target-side154data path of an NVMe-oF target on the BlueField hardware: presenting155`doca_sta_subsystem` targets (NQN + namespaces) backed by local156NVMe-PCI disks (`doca_sta_be`) to remote initiators over RDMA. The157skill is *not* for NVIDIA developers contributing to DOCA STA158itself, and it is *not* for initiator/host-side NVMe stacks.159160**Language scope.** DOCA STA ships as a C library with161`pkg-config` module name `doca-sta`. DOCA STA ships **no public162samples** — the public packages install no163`samples/doca_sta/` directory — so the worked examples in164the tasks companion build against the public headers directly rather than165modify a shipped sample. C and C++ consumers are the canonical166case. Other-language167consumers (Rust, Go, Python, …) consume the same `*.so` through168FFI or language-specific bindings; the skill's contribution in169that case is to keep the target-model, lifecycle,170capability-discovery, queue-pair shape, substrate-dependency,171and error-taxonomy guidance language-neutral, and to route the172agent to the public C ABI as the authoritative surface that any173wrapper will eventually call.174175## When to load this skill176177Load this skill when the user is doing hands-on DOCA STA work,178in any language. Concretely:179180- Initializing a `doca_sta` instance on a `doca_dev` opened181 against a BlueField PF / SF and configuring the NVMe-oF182 target subsystems before `doca_ctx_start()`.183- Defining target resources — `doca_sta_subsystem` (NQN +184 namespaces) and `doca_sta_be` backend controllers (local185 NVMe-PCI disks) — and accepting NVMe-oF connections (admin186 queue plus N I/O queues per connection) on the target side as187 remote initiators connect over RDMA CM.188- Reading or setting STA properties via the `doca_sta_set_*`189 family, checking device support via `doca_sta_cap_is_supported`,190 and querying sizing limits via the `doca_sta_get_max_*` family191 (max I/O queue depth, max number of queue pairs, max I/O size,192 max subsystems, max namespaces per subsystem, max backends).193- Wiring the **NVMe-over-RDMA** transport — STA's only transport;194 it lands on the `doca-rdma` substrate and uses RDMA CM for195 connection establishment — for the target's I/O queues.196- Wiring DOCA Flow rules so that NVMe-oF traffic actually197 reaches the STA-managed queues — the steering boundary is198 `doca-flow`, not `doca-sta`.199- Debugging a `DOCA_ERROR_*` returned from a STA call (lifecycle200 vs. capability vs. transport-layer I/O failure vs.201 driver-below) and the per-queue events on the DOCA Core202 progress engine.203- Designing or extending non-C bindings (Rust, Go, Python, …)204 that wrap the DOCA STA C ABI — for the lifecycle, queue-pair,205 cap-query, and substrate-dependency rules the wrapper must206 honor.207208Do **not** load this skill for general DOCA orientation, install209of DOCA itself, raw RDMA data movement (use210`doca-rdma/SKILL.md`), raw packet I/O on211Ethernet queues (use `doca-eth/SKILL.md`),212flow-rule programming (use `doca-flow/SKILL.md`),213or initiator/host-side NVMe stack development214(SPDK or kernel-nvme own that, not this skill). For DOCA215documentation orientation, use216`doca-public-knowledge-map/SKILL.md`.217218## What this skill provides219220This is a **thin loader**. The body keeps only the orientation221needed to pick the right next file. The substantive STA-specific222material lives in two companion files:223224- the capabilities companion — what DOCA STA can express on this225 version: the target object model (`doca_sta_subsystem` /226 namespaces / `doca_sta_be` backend NVMe-PCI disks),227 the NVMe queue-pair shape (admin queue + I/O queues over RDMA),228 the RDMA-only transport,229 the capability-query surface (`doca_sta_cap_is_supported` plus230 the `doca_sta_get_max_*` sizing queries), the STA231 error taxonomy (mapped onto the cross-library `DOCA_ERROR_*`232 set), the observability surface (per-queue progress engine233 events, capability snapshots), and the safety policy that234 gates substrate-library, permission, and steering235 preconditions.236- the tasks companion — step-by-step workflows for the six in-scope237 STA verbs: `configure`, `modify`, `build`, `run`, `test`,238 `debug`. Plus a `Deferred task verbs` block that points239 out-of-scope questions at the right next skill, and a240 `Command appendix` of the recurring commands the agent241 reaches for.242243The skill assumes a BlueField (with DOCA installed at the244standard location) plus a remote NVMe-oF initiator reachable on245the fabric to connect into the accelerated target, and one or246more local NVMe-PCI disks to back the target's namespaces. It247does not cover installing248DOCA — that path goes through249`doca-setup/SKILL.md`. It does not cover250initiator/host-side NVMe stacks (SPDK `bdev_nvme`, kernel `nvme`251host) or NVMe protocol semantics above the accelerated target252data path — those are out of scope.253254## What this skill deliberately does not ship255256This skill is **agent guidance**, not a samples or templates257bundle. To keep the boundary clean, it deliberately does not258contain — and pull requests should not add:259260- **Pre-written DOCA STA application source code, in any261 language.** DOCA STA ships **no public samples** — there is no262 `/opt/mellanox/doca/samples/doca_sta/` directory. The authoritative surface is the public headers under263 $(pkg-config --variable=includedir doca-common) plus the public264 DOCA STA guide; the agent builds against those directly rather265 than modifying a shipped sample, per the266 the tasks companion workflow.267- **Initiator/host-side NVMe stack glue.** SPDK `bdev_nvme`,268 the kernel `nvme` host, and any initiator-side NVMe stack are269 upstream projects out of scope for this skill — DOCA STA is270 target-side acceleration, not an initiator transport provider.271- **Standalone build manifests** (`meson.build`, `CMakeLists.txt`,272 `Cargo.toml`, …) parked inside the skill. The agent273 constructs the build manifest *in the user's project274 directory* against the user's installed DOCA, where275 `pkg-config --modversion doca-sta` is the source of truth.276- **A `samples/`, `bindings/`, or `reference/` subtree** of any277 kind. A mock or incomplete artifact in this skill's tree,278 even one labeled "reference", is misleading: users will279 read it as buildable.280281## Loading order2822831. Read this skill first to confirm the user's question284 is in scope.2852. **For the STA capability matrix, the target object model,286 queue-pair shape, RDMA-only transport, capability-query287 rules, error taxonomy, observability, and safety policy,288 see the capabilities companion.**2893. **For step-by-step workflows — configure, modify, build,290 run, test, debug — see the tasks companion.**291292Both companion files cross-link to each other,293`doca-version/SKILL.md` for the canonical294version-handling rules,295`doca-rdma/SKILL.md` for the RDMA substrate296that NVMe-over-RDMA transport lands on,297`doca-flow/SKILL.md` for the steering rules298that direct NVMe traffic to STA-managed queues, and299`doca-public-knowledge-map/SKILL.md`300whenever the right answer is "look it up in the public docs or301the installed package layout" rather than "STA-specific302guidance".303304## Related skills305306- `doca-public-knowledge-map/SKILL.md` —307 the routing table for every public DOCA documentation source308 and the on-disk layout of an installed DOCA package. The309 STA URL slug is `DOCA-STA`.310- `doca-setup/SKILL.md` — env preparation,311 install verification, BlueField mode checks, and the312 permission / group-membership requirements for opening a313 `doca_dev`. This skill assumes its preconditions are314 satisfied.315- `doca-version/SKILL.md` — canonical316 DOCA version-handling rules. This skill's `## Version317 compatibility` cross-links the four-way match rule and adds318 only the STA-specific overlay (STA target-acceleration319 availability windows, NVMe-oF feature-set device-conditional320 support).321- `doca-structured-tools-contract/SKILL.md` —322 the bundle's structured-tools precedence rule (detect /323 prefer / fall back / report). The Command appendix in324 the tasks companion honors this contract.325- `doca-programming-guide/SKILL.md` —326 general DOCA programming patterns shared by every library:327 the canonical `pkg-config` + meson build pattern, the328 universal modify-a-shipped-sample first-app workflow, the329 universal lifecycle, the cross-library `DOCA_ERROR_*`330 taxonomy, and the program-side debug order. This skill331 layers STA specifics on top.332- `doca-rdma/SKILL.md` — the RDMA substrate333 that NVMe-over-RDMA transport lands on. STA hides most of334 the RDMA queue-pair details from the consumer, but the user335 still needs `doca-rdma` linked in and the device's RDMA336 capabilities discoverable for the NVMe-over-RDMA path to337 work.338- `doca-eth/SKILL.md` — the queue-pair339 shape that STA's per-connection queue model echoes. Reach340 here if the user is asking general questions about how341 DOCA exposes queue-pairs that don't have an STA-specific342 answer.343- `doca-flow/SKILL.md` — the steering344 surface that decides which NVMe-oF packets land on which345 STA-managed queue. DOCA STA does *not* program steering346 itself; an NVMe-oF target whose connections never come up347 is often a missing or wrong Flow rule, not a STA bug.348- `doca-debug/SKILL.md` — the349 cross-cutting debug ladder (install / version / build /350 link / runtime / program / driver). STA-specific debug351 (transport-type mismatches, queue-depth oversize,352 IO-failed transport errors) overlays on top of that353 ladder.