NVIDIA-NeMo/Gym is a library and infrastructure for evaluating and improving models and agents inside environments, where each environment defines tasks, agent interaction, verification, and execution state. It is for teams running reproducible evaluations or training at scale across settings such as code execution, tool calling, and sandboxes, and the catalogue entries provide skills and instructions for working with it.
Borrowing it
Nothing to install: this file belongs to NVIDIA-NeMo/Gym. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/NVIDIA-NeMo/Gym/main/.agents/skills/nemo-gym-blade-analysis/SKILL.mdgit clone --depth 1 https://github.com/NVIDIA-NeMo/GymWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/nvidia-nemo/gym/nemo-gym-blade-analysis)<a href="https://agentmods.dev/skills/nvidia-nemo/gym/nemo-gym-blade-analysis"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/gym/nemo-gym-blade-analysis/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/nvidia-nemo/gym/nemo-gym-blade-analysis"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/gym/nemo-gym-blade-analysis.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00106 | $0.01713 |
| Opus 5.5 | $0.00042 | $0.00685 |
| Sonnet 5.5 | $0.00021 | $0.00343 |
| Haiku 4.5 | $0.00011 | $0.00171 |
Grade A, and why
nemo-gym-blade-analysis scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 183 lines — stays where its author put it; the contents beside it link to each section on GitHub.
NeMo Gym BLADE Analysis
Invocation Check
Use this skill when the user wants to turn NeMo Gym rollout outputs into an analysis report, benchmark card, model comparison, benchmark-improvement recommendation, or BLADE-ready benchmark package.
Load references/blade-benchmark-build-guide.md when the user asks how to
build, validate, submit, or review a BLADE benchmark or asks whether a benchmark
has all required BLADE deliverables.
Use the bundled public helper at scripts/blade_toolkit.py for package
validation, draft anchor-fact extraction, shallow baseline generation, and local
calibration when external BLADE tooling is not available in the target
repository.
Do not load benchmark-specific examples by default. Load
references/cvdp-report-example.md only when the user explicitly asks for a
CVDP example, the original CVDP report layout, or this optional reference, or
when the agent is confused about the goal and needs one concrete example to
re-anchor on what a BLADE-style report is supposed to look like.
Nemotron-only golden analysis artifacts are available under
references/nemotron-analysis-artifacts/ as original-CVDP example artifacts.
Load them only when the user explicitly asks to study an example completed
BLADE-style report, asks for CVDP artifacts, or the agent is confused about the
goal and needs a concrete completed example. Do not load those files by default.
Inputs To Gather
Start by identifying the artifact set:
- rollout JSONL from
ng_collect_rollouts - aggregate metrics JSON, if present
- reward profile JSONL from
ng_reward_profile, if present - benchmark-specific report directory, if present
- optional golden analysis artifacts, if the user asks to compare against a curated report
- config paths, agent name, model name, repeat count, and sampling settings
- source dataset metadata, license, and known redaction limits
If artifacts are missing, state which claims cannot be supported rather than filling gaps from memory.
If the task is benchmark construction rather than report analysis, first build
an inventory of BLADE deliverables: analysis skill, rollout data, and golden
report packages with metrics and anchor facts. Current BLADE scoring is handled
by the universal blade-judge; benchmark-local judge utilities are optional
pre-checks, not required deliverables. Missing deliverables are blocking work
items, not footnotes.
What ships with it
8 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- agents/openai.yaml 257 B
- references/blade-benchmark-build-guide.md 13 KB
- references/cvdp-report-example.md 4.7 KB
- references/nemotron-analysis-artifacts/nemotron_3_super_anchor_facts.json 8.4 KB
- references/nemotron-analysis-artifacts/nemotron_3_super_golden_report_metrics.json 517 B
- references/nemotron-analysis-artifacts/nemotron-3-super-golden-report.md 28 KB
- references/nemotron-analysis-artifacts/README.md 669 B
- scripts/blade_toolkit.py 24 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 183 lines · 106 tokens per session scan A a7ead8dca30c
nemo-gym-blade-analysis is a skill published in the GitHub repository NVIDIA-NeMo/Gym (1,320 stars, last pushed today), licensed Apache-2.0. It adds 106 tokens to every session and 1,713 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
nemo-gym-blade-analysis compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| nemo-gym-blade-analysisNVIDIA-NeMo/Gym | Skill | 1.3k | today | Apache-2.0 | A | 106 | Codex |
| dfam-checkearthtojake/text-to-cad | Skill | 26k | yesterday | MIT | A | 117 | Claude Code, Codex |
| tooluniverse-variant-analysismims-harvard/ToolUniverse | Skill | 1.8k | yesterday | Apache-2.0 | A | 317 | Claude Code |
| tooluniverse-phylogeneticsmims-harvard/ToolUniverse | Skill | 1.8k | yesterday | Apache-2.0 | A | 155 | Claude Code |
| tooluniverse-comparative-genomicsmims-harvard/ToolUniverse | Skill | 1.8k | yesterday | Apache-2.0 | A | 90 | Claude Code |
| tooluniverse-gene-enrichmentmims-harvard/ToolUniverse | Skill | 1.8k | yesterday | Apache-2.0 | A | 102 | Claude Code |
| tooluniverse-systems-biologymims-harvard/ToolUniverse | Skill | 1.8k | yesterday | Apache-2.0 | A | 72 | Claude Code |