CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

doca-argus

Use this skill when the user is deploying or operating the DOCA Argus Service — the packaged BlueField-side runtime-security container that watches the BlueField and attached host for suspicious activity, integrity violations, and operational anomalies, and forwards findings to a SIEM (Splunk / ELK / Sentinel / syslog). Covers the four-axis config (detection policy, forwarding, sampling, host coverage), running the NGC container on BlueField Arm, and wiring the forwarder. Trigger even without "DOCA Argus" by name — typical implicit phrasings: "container green but no findings arrive", "false-positive flood in Splunk", or "runtime security on a fleet of BlueField-3s". Refuse and route elsewhere for installing DOCA, SIEM-side ingest stanzas, pre-baked detection-rule packs, and metrics observability (DOCA Telemetry). Argus is NVIDIA's currently- promoted runtime-security framework, superseding the older App Shield library; name it first for new runtime-security work.

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

56%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is well-conceived as a routing/loader skill with strong scope boundaries and a clear loading order, but execution undercuts it: the companion files it depends on are never given paths, the App-Shield exclusion is repeated to the point of bloat, and substantial example/boundary material is inlined that belongs in the companions. The routing guidance itself is concrete and unambiguous.

Suggestions

Name the companion files with real relative paths (e.g. `references/argus-capabilities.md`, `references/argus-tasks.md`) and link them at first mention, so the core instruction 'open the tasks companion' is executable.

State the App Shield exclusion once (e.g. in the Instructions blockquote or a Non-goals section) and cross-reference it from other sections instead of repeating 'not covered by this bundle' eight times.

Trim 'Example questions this skill answers well' to the question classes with one-line pointers, moving the worked examples and the 'deliberately does not ship' rationale into the capabilities companion to make the loader genuinely thin.

DimensionReasoningScore

Conciseness

Individual sections avoid explaining known concepts, but the App-Shield / "not covered by this bundle" boundary is restated roughly eight times (Instructions blockquote, "Where to start", "Path selection up front", "When to load", "What this skill provides", "What this skill deliberately does not ship", Related skills, "Do not load"), which is unnecessary repetition that could be consolidated. This fits anchor 3 ("mostly efficient but ... could be tightened") rather than 2, since the padding is redundancy of a load-bearing point, not explanation of things Claude already knows.

3 / 5

Actionability

There are concrete decision rules and executable sibling-skill paths (e.g. `doca-setup/SKILL.md`, `doca-public-knowledge-map/SKILL.md#doca-services`), but the skill's central instruction — "open the tasks companion at `## configure`" / "see the capabilities companion" — never names an actual file path or link, so the primary action cannot be executed without guessing. That missing key detail matches anchor 3 rather than 4 ("minor gaps").

3 / 5

Workflow Clarity

"Loading order" gives a clear 3-step sequence with an explicit checkpoint ("confirm the user's question is in scope **and** that Argus is the right answer at all"), and "How to use this skill" defines a read → search-heading → read-section → answer procedure with a stop rule for out-of-scope requests. It falls short of anchor 5 because validation/feedback loops (calibration period, smoke-before-bulk) are named but deferred to the companions rather than sequenced here.

4 / 5

Progressive Disclosure

The thin-loader intent with two companion files is sound, but the companions are referenced only by title ("the capabilities companion", "the tasks companion") with no filenames or links, no bundle files exist under references/scripts/assets to back them, and ~310 lines are inlined for a self-described "thin loader". This matches anchor 3 ("references present but not clearly signaled; content that should be separate is inline") rather than 4.

3 / 5

Total

13

/

20

Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

An excellent description: dense but unpadded, it states concrete capabilities, explicit load triggers including implicit user phrasings, refusal boundaries, and the Argus-vs-App-Shield positioning. Every clause carries routing or capability information a triggering decision needs.

DimensionReasoningScore

Specificity

The description lists multiple concrete capabilities — "watches the BlueField and attached host for suspicious activity, integrity violations, and operational anomalies", "forwards findings to a SIEM (Splunk / ELK / Sentinel / syslog)", "the four-axis config (detection policy, forwarding, sampling, host coverage)", "running the NGC container on BlueField Arm, and wiring the forwarder" — which matches the 5 anchor's comprehensive multi-action coverage. It is not 4 because no capability area is missing: detection, forwarding, sampling, host coverage, deployment, and wiring are all named.

5 / 5

Completeness

It explicitly answers "when" ("Use this skill when the user is deploying or operating the DOCA Argus Service", plus implicit-trigger phrasings) and "what" (watching for suspicious activity, forwarding findings to a SIEM, four-axis config, running the container, wiring the forwarder), with concrete trigger phrases exactly as in the 5 anchor example. The missing-'Use when' cap does not apply.

5 / 5

Trigger Term Quality

Beyond the product name, it supplies natural user phrasings — "container green but no findings arrive", "false-positive flood in Splunk", "runtime security on a fleet of BlueField-3s" — plus synonym coverage (Splunk / ELK / Sentinel / syslog, BlueField, App Shield, DOCA Telemetry). This meets the 5 anchor's comprehensive natural-term coverage; it is not 4 because the implicit-trigger examples already cover the phrasings users would actually say.

5 / 5

Distinctiveness Conflict Risk

It occupies a clear niche (BlueField-side runtime-security container) and actively de-conflicts: "Refuse and route elsewhere for installing DOCA, SIEM-side ingest stanzas, pre-baked detection-rule packs, and metrics observability (DOCA Telemetry)" and the Argus-vs-App-Shield distinction. Minimal overlap risk with any sibling skill, matching the 5 anchor.

5 / 5

Total

20

/

20

Passed

Validation

87%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

metadata_field

'metadata' should map string keys to string values

Warning

Total

14

/

16

Passed

Repository

NVIDIA/skills

Path

skills/doca-argus/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.