{"id":"d694db29-4fe5-46de-89c5-fc1f6c0c71a4","slug":"doca-caps","kind":"skill","name":"doca-caps","summary":"Use this skill when the user wants to invoke the read-only doca_caps CLI to ask what DOCA sees on this host — listing DOCA devices and PCIe addresses, listing representor devices, asking which DOCA li","owner":"NVIDIA","repoUrl":"https://github.com/NVIDIA/skills/tree/HEAD/skills/doca-caps","license":"Apache-2.0","licenseTier":"permissive","trust":"official","stars":3586,"match":null,"fewStars":false,"format":"skill","createdAt":null,"addedAt":"2026-10-14T05:42:58.434Z","logoUrl":"https://avatars.githubusercontent.com/u/1728152?v=4","linkOnly":false,"collection":{"name":"skills","repo":"nvidia/skills"},"skillPath":"skills/doca-caps","categories":["developer-tools"],"tags":["agentic-ai","ai-agents","claude-code","codex","coding-agents","cuda","developer-tools","llm"],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-14T05:42:58.418Z","sample":false,"url":"https://conduits.dev/tools/doca-caps","language":"Python","transports":[],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"> Use this skill when the user wants to invoke the read-only doca_caps CLI to ask what DOCA sees on this host — listing DOCA devices and PCIe addresses, listing representor devices, asking which DOCA libraries are available on the current OS, checking per-device per-library capabilities, scoping output to a specific PCIe address, or capturing a side-effect-free capability snapshot for a debug session or install smoke-test. Trigger even when the user does not explicitly mention \"doca_caps\" or \"capabilities print tool\" — typical implicit phrasings include \"what does DOCA actually see on this box\", \"is my BlueField PF visible to DOCA\", \"is Flow available on my RHEL host\", \"enumerate VF representors for pf0\", \"doca_caps: command not found\", or \"empty output for RDMA, is the tool broken\". Refuse and route elsewhere for DOCA installation, library-internal capability matrices (Flow pipe creation, RDMA verbs features), streaming telemetry / DTS, or modifying the shipped binary — those belong to other skills.\n\n# DOCA Capabilities Print Tool (`doca_caps`)\n\n**Where to start:** This is a tool skill for invoking `doca_caps`,\na side-effect-free CLI. Open [`TASKS.md`](TASKS.md) and start at\n[`## run`](TASKS.md#run) for the documented invocations, or\n[`## test`](TASKS.md#test) when using `doca_caps` as an install\nsmoke-test. Open [`CAPABILITIES.md`](CAPABILITIES.md) when the\nquestion is *what kinds of capability families `doca_caps` reports*.\nIf DOCA is not installed yet, route to\n[`doca-setup`](../../doca-setup/SKILL.md) first.\n\n## Example questions this skill answers well\n\nThe CLASSES of `doca_caps` questions this skill is built to answer,\neach with one worked example. The class is the load-bearing piece;\nthe worked example is one instance.\n\n- **\"What DOCA devices does this host see?\"** — worked example: *\"is\n  my BlueField PF visible to DOCA\"*. Answered by the device\n  enumeration in\n  [`CAPABILITIES.md ## Capabilities and modes`](CAPABILITIES.md#capabilities-and-modes)\n  + the `--list-devs` invocation in\n  [`TASKS.md ## run`](TASKS.md#run).\n- **\"Which DOCA libraries are available on this OS?\"** — worked\n  example: *\"is Flow available on my RHEL host\"*. Answered by the\n  library-availability surface in\n  [`CAPABILITIES.md ## Capabilities and modes`](CAPABILITIES.md#capabilities-and-modes)\n  + the library-listing invocation in\n  [`TASKS.md ## run`](TASKS.md#run).\n- **\"Does this device support library X capability Y?\"** — worked\n  example: *\"does this device support Flow hairpin?\"*. Answered by\n  the per-device per-library capability matrix in\n  [`CAPABILITIES.md ## Capabilities and modes`](CAPABILITIES.md#capabilities-and-modes)\n  + the `--pci-addr`-scoped invocation in\n  [`TASKS.md ## run`](TASKS.md#run).\n- **\"What representors are visible to DOCA?\"** — worked example:\n  *\"enumerate VF representors for pf0\"*. Answered by the representor\n  enumeration in\n  [`CAPABILITIES.md ## Capabilities and modes`](CAPABILITIES.md#capabilities-and-modes)\n  + the `--list-rep-devs` invocation in\n  [`TASKS.md ## run`](TASKS.md#run).\n- **\"I want a snapshot of state to attach to my debug session.\"** —\n  worked example: *\"save device + library + capability output to a\n  file\"*. Answered by the snapshot workflow in\n  [`TASKS.md ## test`](TASKS.md#test) and consumed by\n  [`doca-debug ## test`](../../doca-debug/TASKS.md#test) step 3\n  (read-only triple) and\n  [`doca-programming-guide ## debug`](../../doca-programming-guide/TASKS.md#debug).\n- **\"`doca_caps` returned nothing for capability Y — what does that\n  mean?\"** — worked example: *\"empty output for RDMA\"*. Answered by\n  the empty-output interpretation rules in\n  [`TASKS.md ## debug`](TASKS.md#debug) +\n  [`CAPABILITIES.md ## Error taxonomy`](CAPABILITIES.md#error-taxonomy).\n\n## Audience\n\nThis skill serves **external operators, developers, and AI agents who\nneed a side-effect-free way to ask \"what does DOCA see on this host?\"**\nbefore doing anything that changes state. Concretely:\n\n- An external developer who installed DOCA (or is using the public NGC\n  DOCA container per [`doca-setup ## no-install`](../../doca-setup/TASKS.md#no-install))\n  and wants to confirm the install can see hardware before writing code.\n- A platform operator deploying a DOCA service who wants a documented,\n  read-only snapshot of *which DOCA libraries this host actually\n  supports* and *which capabilities each DOCA device offers*.\n- An AI agent producing a *capability snapshot* artifact during the\n  documented setup or programming-guide debug procedures (it's listed\n  as the canonical first step in\n  [`doca-setup ## test`](../../doca-setup/TASKS.md#test) and\n  [`doca-programming-guide ## debug`](../../doca-programming-guide/TASKS.md#debug)).\n\nIt is **not** for users debugging `doca_caps` itself, and **not** a\nsubstitute for the live public Capabilities Print Tool guide.\n\n`doca_caps` is shipped as a **tool** (a single CLI binary), not a\nlibrary you link against. The skill uses the same `kind: tool`\nthree-file shape as the rest of the bundle so the agent's task-verb\ncontract (`configure / build / modify / run / test / debug`) is uniform\nacross libraries, services, and tools — even when individual verbs\ncollapse to a routing stub for a shipped read-only binary.\n\n## When to load this skill\n\nLoad this skill when the user is — or the agent needs to — invoke\n`doca_caps` on a real host with DOCA installed (or inside the public\nNGC DOCA container). Concretely:\n\n- Running `doca_caps --list-devs` to enumerate DOCA devices.\n- Running `doca_caps --list-rep-devs` to enumerate representor\n  devices.\n- Scoping output to a specific PCIe address with `--pci-addr`.\n- Listing the DOCA libraries the install reports as available on the\n  current OS.\n- Listing the available DOCA logger names.\n- Capturing a documented, side-effect-free **capability snapshot** as\n  prerequisite evidence for later `## debug` workflows.\n\nDo **not** load this skill for general DOCA orientation, library API\nwork, or installation. For those, use\n[`doca-public-knowledge-map`](../../doca-public-knowledge-map/SKILL.md),\nthe matching `libs/<library>` skill, or\n[`doca-setup`](../../doca-setup/SKILL.md).\n\n## What this skill provides\n\nThis is a **thin loader**. Substantive material lives in two companion\nfiles:\n\n- `CAPABILITIES.md` — what `doca_caps` reports (the five documented\n  capability families: devices, representors, libraries, library\n  capabilities, loggers), version availability and execution\n  environment, the tool's narrow error surface, its observability role\n  inside other skills' workflows, and its read-only safety posture.\n- `TASKS.md` — step-by-step workflows for the in-scope task verbs:\n  `configure` (route to install), `build` (route to install), `modify`\n  (refuse), `run` (the documented invocations), `test` (capability\n  snapshot as install smoke-test), `debug` (what to do when the tool\n  reports nothing or fails), plus a `Deferred task verbs` block routing\n  out-of-scope questions.\n\nThe skill assumes a host where DOCA is already installed (or the public\nNGC DOCA container is running) and the operator has whatever\npermissions the public guide requires for `doca_caps` to enumerate\ndevices on their platform.\n\n## What this skill deliberately does not ship\n\nThis skill is **agent guidance**, not a samples or scripts bundle. To\nkeep the boundary clean, it deliberately does not contain — and pull\nrequests should not add:\n\n- **Pre-baked example output.** Output is install- and\n  hardware-specific. A captured example pinned to one platform and\n  one DOCA version misleads operators on a different platform / version.\n- **Wrappers, parsers, or scripts** in any language that consume\n  `doca_caps` output. The output format is documented; if a user\n  wants to script against it, the right answer is \"read the live\n  guide, write the parser against your installed version\".\n- **A `samples/` or `reference/` subtree.** This is a thin loader for\n  a documented CLI; substantive material lives on the public page and\n  in `--help`.\n\n## Loading order\n\n1. Read this `SKILL.md` first to confirm the user's question is in\n   scope (the user actually wants to invoke `doca_caps`, not learn\n   about DOCA in general).\n2. **For what `doca_caps` reports, version availability, error\n   surface, and safety posture, see [CAPABILITIES.md](CAPABILITIES.md).**\n3. **For the documented invocations and the capability-snapshot\n   workflow — `configure`, `build`, `modify`, `run`, `test`, `debug` —\n   see [TASKS.md](TASKS.md).**\n\n## Related skills\n\n- [`doca-public-knowledge-map`](../../doca-public-knowledge-map/SKILL.md)\n  — routing to the public Capabilities Print Tool guide and the rest\n  of the public DOCA documentation set.\n- [`doca-setup`](../../doca-setup/SKILL.md) — env preparation, install\n  verification (`doca_caps` is the canonical first step there), and\n  the *I have no install yet* path with the public NGC DOCA container.\n- [`doca-programming-guide`](../../doca-programming-guide/SKILL.md) —\n  cross-library programming patterns, including the `## debug`\n  procedure where the saved `doca_caps` snapshot is consumed.\n- The matching `libs/<library>` skill — for fine-grained,\n  library-specific capability questions that go beyond what\n  `doca_caps` exposes."}