Nemo Relay Get Started

NVIDIAOfficialCreated by NVIDIA Corporation and Affiliates

Skill
Tool integration
Automation and integrations
Apache-2.0
Medium risk

From the author

Use this skill when first-time NeMo Relay users want help selecting and following an appropriate quick-start guide or verifying their first working integration. Do not use it for production rollout or broad integration development.

— the description by NVIDIA Corporation and Affiliates

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

The nemo-relay-get-started skill assists new users in navigating the NeMo Relay ecosystem by identifying the correct quick-start guide based on their specific runtime, framework, or application environment. It focuses on providing a structured, evidence-based path to verify a successful initial integration. The skill emphasizes adherence to official documentation, version compatibility, and the use of guide-specific verification steps, ensuring that users establish a working proof-of-concept without unnecessary configuration or production-level complexity.

What it does

Identifies the appropriate quick-start guide based on user goals and environment.
Routes users to specific trial paths for CLI, agent runtimes, frameworks, or manual integrations.
Ensures commands and configurations are derived directly from the selected guide's current version.
Facilitates guide-based verification to confirm successful integration.
Provides guidance on transitioning from an initial trial to further plugin configuration.
Maintains project continuity by preserving existing settings and workflows.

When to use it

When a user is new to NeMo Relay and needs help selecting a starting point.
When a user needs to verify their first working integration with NeMo Relay.
When a user requires assistance following a specific quick-start guide for their environment.
When a user needs to confirm that their initial setup meets the guide's success criteria.

Advantages

Reduces setup friction by providing a clear, guided path for initial integration.
Ensures technical accuracy by relying on version-matched, official documentation.
Prevents configuration errors by adhering to guide-specific prerequisites and verification steps.
Maintains the integrity of the user's existing project environment.
Provides a clear progression path from initial proof to broader instrumentation.

Usage examples

“I am new to NeMo Relay and want to try it with LangChain. How do I get started?”: The agent identifies the LangChain quick-start guide, verifies the installed release compatibility, and walks the user through the specific attachment and verification steps for LangChain.
“I need to set up NeMo Relay for my Hermes runtime. What is the first step?”: The agent routes the user to the Hermes-specific quick-start guide, ensuring the user follows the native integration path rather than installing unnecessary wrappers.

Install

Download the skill as a ZIP:
https://mp-staging.autogineer.com/agent-skills/nvidia/nemo-relay-get-started/download

Unzip it into the cross-client skills folder of your project:

mkdir -p .agents/skills && unzip nemo-relay-get-started.zip -d .agents/skills

Documentation by NVIDIA Corporation and Affiliates

Get Started With NeMo Relay

Help the user choose and follow the maintained quick-start guide that fits their existing runtime, framework, plugin, or application. The guide owns the implementation; this skill owns selection, continuity, and evidence.

Select The Guide

Read the current quick-start index. Use the user's stated goal and existing environment to find the smallest applicable path. Preserve a target established in conversation or installation.

When the boundary is unclear, ask one concise question before choosing. Do not infer that every generic "try Relay" request means a coding-agent demo, or introduce a framework just to demonstrate Relay.

Use these references only for target-specific routing and cautions:

Established boundary	Reference
Local coding agent or CLI	CLI Trial
Native agent runtime or managed plugin	Agent Runtime Trial
Maintained framework integration	Framework Trial
Switchyard routing	Switchyard Trial
Application-owned language calls	Language Trial

Follow The Current Guide

Read the selected guide before proposing commands or configuration, including its prerequisites, sample files, run instructions, verification, and limitations. Use documentation compatible with the installed stable release. The current published guide is the normal source; use an explicitly requested pending PR or other supplied revision as such, without treating it as published support. Resolve the documentation version instead of assuming an unversioned URL is current. When rendered documentation is inaccessible or omits the selected path, use the official source at the matching release tag, or the development checkout for an explicitly requested development preview. A failed HTTP request alone does not prove that a guide was removed; check its index and source.

Keep commands and referenced sample files from the same guide revision. If working from source, resolve its referenced files in that checkout. Do not reconstruct the recipe from this skill, an older session, or a benchmark. If the guide cannot be read or its sample files are unavailable, state the gap rather than inventing instructions.

For missing dependencies, use nemo-relay-install with the already-selected target. Its default is the latest published stable release, subject to explicit user or project constraints. Preserve a successful install and continue at the next appropriate guide step; do not repeat target selection.

A guide's example pin is not an instruction to downgrade or use an unpublished release. If the guide and installed release are incompatible, resolve that mismatch using current release documentation before proceeding.

Execute And Verify

Use Guide-Based First-Value Checks while adapting only what the user's existing environment requires.

Explain the guide's attachment boundary and proposed changes. Preserve the project's manager, unrelated settings, and existing artifacts. Proceed within existing authorization, honoring requests to preview or stop before a live run.

Follow the selected guide's configuration, execution, shutdown, and verification steps. Derive file locations, schemas, commands, prompts, lifecycle expectations, and output formats from that guide rather than imposing a shared recipe. Run only the chosen path and report the evidence it actually establishes.

Successful application output alone is not proof that Relay captured the work. Use fresh evidence and the guide's own success criteria. Do not demand model credentials, a tool call, ATIF, or another exporter when that guide does not.

After The First Proof

Stop the initial trial when the guide's success checks pass. Explain the observed value and Relay's progression: attach once, then extend supported behavior through plugin configuration.

Resolve plugin and integration documentation for the installed Relay release, including any explicit version constraint, before suggesting one next plugin for the user's goal. Use the stable plugin overview only after confirming compatibility with that release; otherwise use the release-matched documentation or official source. Do not repeat Observability setup if the guide already established it, or offer capabilities unsupported by the selected integration.

Hand broader instrumentation to nemo-relay-instrument-calls, unsupported framework work to nemo-relay-integrate-upstream, and additional export to nemo-relay-plugin-observability. Persistent host setup belongs to nemo-relay-install, including its applicable Codex Desktop continuity gate. Production deployment and multiple new behaviors belong to follow-up work.

Files

ATTRIBUTION.mdAdded by AutoGineer951 B
BENCHMARK.md11.3 KiB
LICENSELicence file added by AutoGineer11.2 KiB
SKILL.md5.3 KiB
evals/evals.json19.1 KiB
references/agent-runtimes-try-now.md1.8 KiB
references/built-in-integrations-try-now.md1.4 KiB
references/cli-try-now.md1.8 KiB
references/first-value.md2.6 KiB
references/manual-language-try-now.md1.5 KiB
references/switchyard-try-now.md1.7 KiB
skill-card.md4.8 KiB
skill.oms.sig5.9 KiB

BENCHMARK.md

Skill Benchmark: nemo-relay-get-started

✅ Overall verdict: PASS — Recommended for publication

Publication Recommendation

Recommended for publication based on the completed evaluation evidence in this report.

Evaluation Metadata

Skill: nemo-relay-get-started
Evaluation date: 2026-10-22
Evaluator version: 1.5.6
Agents: Claude Code (aws/anthropic/bedrock-claude-sonnet-5-5), Codex (openai/openai/gpt-5.5)
Tasks: 22 evaluation tasks (21 positive, 1 negative)
Dataset digest: sha256:477cc9bd7b45256aa6d8f837938c05e6d79c539ed8bce8bf1908e266f61f3e9f (skill-evaluator-dataset-snapshot/1)
Attempts per task: 1
Environment: k8s-sandbox
Tier 2 evidence: required for publication
Tier 3 evidence: required for publication

Each task attempt ran in its own isolated sandbox pod.

What This Report Answers

The three-tier evaluation checks whether the skill:

is safe to use;
produces correct answers;
is discovered and activated when needed;
helps the agent complete the user's goal and expected workflow; and
avoids wasted skill and tool usage.

Results at a Glance

Measure	Claude Code (Baseline → Skill Uplift)	Codex (Baseline → Skill Uplift)
Overall	81.0% — baseline ran, but no comparable score was available; uplift unavailable	71.3% — baseline ran, but no comparable score was available; uplift unavailable
Security	86.4% → 86.4% (±0.0 points)	79.6% → 70.5% (-9.1 points)
Correctness	33.6% → 84.6% (+51.0 points)	56.4% → 85.5% (+29.1 points)
Discoverability	92.7% — baseline ran, but no comparable score was available; uplift unavailable	83.8% — baseline ran, but no comparable score was available; uplift unavailable
Effectiveness	34.4% → 62.9% (+28.5 points)	38.7% → 55.5% (+16.8 points)
Efficiency	78.5% — baseline ran, but no comparable score was available; uplift unavailable	61.0% — baseline ran, but no comparable score was available; uplift unavailable

How to read this table: baseline is the same task attempted without the target skill. Scores are rounded to one decimal; threshold-adjacent values use additional precision so their displayed band matches the verdict. Uplift is derived from those displayed scores and shown in percentage points.

Example: 47.0% → 92.0% (+45.0 points) means the skill-assisted run scored 92.0%, 45.0 percentage points above its 47.0% no-skill baseline.

A partial dimension was calculated from only the available configured signals; review the detailed report before relying on it.

Token Usage

Actual Tier 3 execution usage is reported for every observed agent/case pair and both conditions.

Agent	Dataset case	With skill	Without skill	Delta	Change	Coverage
claude-code	All cases	8,269,624	3,485,549	+4,784,075	+137.25%	skill 22/22; base 22/22
claude-code	nemo-relay-get-started-001	241,424	89,531	+151,893	+169.65%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-002	465,266	89,838	+375,428	+417.89%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-003	1,463,446	153,853	+1,309,593	+851.20%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-004	89,411	29,043	+60,368	+207.86%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-005	238,131	92,203	+145,928	+158.27%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-006	203,743	121,315	+82,428	+67.95%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-007	471,803	89,896	+381,907	+424.83%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-008	460,852	89,464	+371,388	+415.13%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-009	130,898	90,247	+40,651	+45.04%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-010	152,580	133,923	+18,657	+13.93%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-011	375,286	90,170	+285,116	+316.20%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-012	30,948	30,704	+244	+0.79%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-013	173,818	89,630	+84,188	+93.93%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-014	581,420	90,133	+491,287	+545.07%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-015	392,726	125,348	+267,378	+213.31%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-016	482,696	121,978	+360,718	+295.72%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-017	389,983	583,370	-193,387	-33.15%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-018	615,449	770,940	-155,491	-20.17%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-019	415,012	121,441	+293,571	+241.74%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-020	130,578	90,211	+40,367	+44.75%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-021	297,366	237,340	+60,026	+25.29%	skill 1/1; base 1/1
claude-code	nemo-relay-get-started-022	466,788	154,971	+311,817	+201.21%	skill 1/1; base 1/1
codex	All cases	12,363,606	8,470,742	+3,892,864	+45.96%	skill 22/22; base 22/22
codex	nemo-relay-get-started-001	34,536	18,302	+16,234	+88.70%	skill 1/1; base 1/1
codex	nemo-relay-get-started-002	786,313	40,813	+745,500	+1826.62%	skill 1/1; base 1/1
codex	nemo-relay-get-started-003	439,518	425,422	+14,096	+3.31%	skill 1/1; base 1/1
codex	nemo-relay-get-started-004	13,297	13,237	+60	+0.45%	skill 1/1; base 1/1
codex	nemo-relay-get-started-005	1,054,171	625,168	+429,003	+68.62%	skill 1/1; base 1/1
codex	nemo-relay-get-started-006	78,490	13,436	+65,054	+484.18%	skill 1/1; base 1/1
codex	nemo-relay-get-started-007	88,674	32,031	+56,643	+176.84%	skill 1/1; base 1/1
codex	nemo-relay-get-started-008	39,197	74,301	-35,104	-47.25%	skill 1/1; base 1/1
codex	nemo-relay-get-started-009	39,604	116,973	-77,369	-66.14%	skill 1/1; base 1/1
codex	nemo-relay-get-started-010	85,988	27,658	+58,330	+210.90%	skill 1/1; base 1/1
codex	nemo-relay-get-started-011	32,923	51,971	-19,048	-36.65%	skill 1/1; base 1/1
codex	nemo-relay-get-started-012	31,298	38,018	-6,720	-17.68%	skill 1/1; base 1/1
codex	nemo-relay-get-started-013	41,136	66,886	-25,750	-38.50%	skill 1/1; base 1/1
codex	nemo-relay-get-started-014	54,303	39,969	+14,334	+35.86%	skill 1/1; base 1/1
codex	nemo-relay-get-started-015	1,781,789	552,045	+1,229,744	+222.76%	skill 1/1; base 1/1
codex	nemo-relay-get-started-016	4,540,882	109,076	+4,431,806	+4063.04%	skill 1/1; base 1/1
codex	nemo-relay-get-started-017	90,050	31,479	+58,571	+186.06%	skill 1/1; base 1/1
codex	nemo-relay-get-started-018	1,220,978	5,786,295	-4,565,317	-78.90%	skill 1/1; base 1/1
codex	nemo-relay-get-started-019	1,669,588	210,266	+1,459,322	+694.04%	skill 1/1; base 1/1
codex	nemo-relay-get-started-020	43,549	13,420	+30,129	+224.51%	skill 1/1; base 1/1
codex	nemo-relay-get-started-021	157,687	120,182	+37,505	+31.21%	skill 1/1; base 1/1
codex	nemo-relay-get-started-022	39,635	63,794	-24,159	-37.87%	skill 1/1; base 1/1
ALL AGENTS	Dataset aggregate	20,633,230	11,956,291	+8,676,939	+72.57%	skill 44/44; base 44/44

Prompt tokens include cached reads, so total tokens are prompt + completion (cached is not added twice). The Efficiency score uses (prompt - cached) + completion. N/A means the relevant trajectory counters were not available; coverage is never estimated.

Tier Status

Tier	Purpose	Status	Evidence
Tier 1	Static validation	PASSED WITH OBSERVATIONS	11 validator(s); 11 finding(s)
Tier 2	Semantic deduplication	PASSED	2 validator(s); 0 finding(s)
Tier 3	Live agent evaluation	PASS	2 agent(s); 22 task(s)

Findings and Observations

<details> <summary>Show detailed findings and successful checks</summary>

MEDIUM QUALITY/quality_correctness: SKILL_SPEC recommended field missing: 'metadata.tags' (skills/nemo-relay-get-started/SKILL.md)
MEDIUM QUALITY/quality_efficiency: Deeply nested references in agent-runtimes-try-now.md (skills/nemo-relay-get-started/SKILL.md)
MEDIUM SCHEMA/body_recommended_section: Missing recommended section: '## Instructions' (skills/nemo-relay-get-started/SKILL.md)
MEDIUM SCHEMA/body_recommended_section: Missing recommended section: '## Examples' (skills/nemo-relay-get-started/SKILL.md)
LOW QUALITY/quality_correctness: No examples provided (skills/nemo-relay-get-started/SKILL.md)
6 additional finding(s) are available in the full evaluation artifacts.

</details>

Scoring Methodology

<details> <summary>Show dimension definitions, source signals, and thresholds</summary>

Dimension	Question	Scored signals
Security	Is it safe to use?	`security` (100%)
Correctness	Is the answer correct?	`accuracy` (100%)
Discoverability	Was the right skill loaded when needed?	`skill_execution` (100%)
Effectiveness	Did the skill help complete the task?	`goal_accuracy` (50%) + `behavior_check` (50%)
Efficiency	Did it avoid wasted tool calls and token usage?	`skill_efficiency` (50%) + `token_efficiency` (50%)

Dimension bands: PASS at 50% or above; NEUTRAL from 40% to below 50%; FAIL below 40%.
Overall Tier 3 lift: PASS at +5 points or more; FAIL at -10 points or less; values between those bands are NEUTRAL.
Overall verdict: PASS only when every configured dimension passes for at least one supported agent. Lift is reported as diagnostic evidence and does not override this gate.
The 50% attempt pass threshold is a separate per-task gate; it is not the dimension pass threshold.
Effectiveness is the equal-weight mean of goal completion (goal_accuracy) and expected workflow adherence (behavior_check).
Efficiency is 50% tool-call productivity (the backward-compatible skill_efficiency wire id) and 50% token_efficiency. Positive-case skill routing is scored under Discoverability, not Efficiency; a negative case without a routing target is N/A. N/A sources are omitted, remaining weights are renormalized, and the dimension is marked partial.

Signals present in this run:

security (Security): unsafe operations, secret leakage, and unauthorized access.
skill_execution (Skill Execution): whether the expected skill was selected, decoys were avoided, and the workflow executed.
skill_efficiency (Tool Productivity): tool-call productivity (legacy wire id; routing is scored under Discoverability).
accuracy (Accuracy): final-answer correctness against the reference answer.
goal_accuracy (Goal Accuracy): whether the user's goal was achieved.
behavior_check (Behavior Check): whether the expected workflow behavior was followed.
token_efficiency (Token Efficiency): actual uncached prompt plus completion usage (50% of Efficiency).

</details>

Freshness

Regenerate this benchmark when the skill, evaluation dataset, target agent/model, evaluator version, environment, or scoring policy changes.

Requirements and compatibility

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Security

What our scanner noticed· Medium risk

Large base64-encoded blobs

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Created by

NVIDIA Corporation and Affiliates

Publisher

NVIDIA (Official)

Source

NVIDIA/skills · skills/nemo-relay-get-started · commit fce3cca

Licence

Apache License 2.0 (redistribution allowed)

LICENSE (added by AutoGineer from LICENSE-APACHE)
ATTRIBUTION.md (written by AutoGineer)

Synced

2026-11-08

AutoGineer is not affiliated with or endorsed by NVIDIA Corporation and Affiliates or NVIDIA.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download nemo-relay-get-started.zip