Skill Claude CodeCodex
Reactive input for NOOA agents — Channel/QueueManager for queued and event-mode input, race() turn dispatch, spawn() background jobs with JobHandle, and the bundled producers (monitor a shell command, cron ticks, one-shot timers, file tails). Use when an agent must react to external input mid-run — user messages, CI…
Skill Claude CodeCodex
Advanced tuning of NOOA strategies — CodeAct prefill (understanding, disabling, custom, pre-ellipsis code), loop guards (maxiterations, retries, text-only stop), truncation tuning (TruncationConfig/CaptureConfig/FormatConfig), code restrictions (RestrictionsConfig), execution environment internals, and PredictStrategy…
Skill Claude CodeCodex
Manage what a NOOA agent sees and remembers — context blocks, event history and queries, history summarization, and persistent memory/storage. Use when pinning information into the system prompt, querying past events, bounding context growth in long conversations, or persisting agent state.
Skill Claude CodeCodex
Intercept and observe NOOA execution — middleware via eventmanager.intercept() (guardrails, input/output transforms, blocking), event observers via eventmanager.on() (react to Task/Error/LLMComplete/turn events), and the InstrumentationHooks protocol for observability backends. Use when adding guardrails, redacting or…
Skill Claude CodeCodex
Let NOOA agents extend themselves — persistent skill libraries the agent writes and hot-reloads (SkillWriting / self.libs, LibraryManager), in-cell helper functions and standalone @strategy sub-calls (MethodWriting), and @slashcommand user actions. Use when an agent should accumulate reusable code across sessions…
Skill Claude CodeCodex
Give a NOOA agent capabilities — methods as tools, built-in tools (ShellTools, TodoManager), MCP servers, agent skills (Skill/TextSkill/SkillRegistry), and multimodal media. Use when adding tools or external integrations to an agent, wiring MCP, or packaging reusable guidance as an agent skill.
Skill Claude CodeCodex
Analyze NOOA execution traces programmatically with the trace-explorer CLI and Python API. Use when debugging why an agent run failed, inspecting LLM turns and code executions, diffing two runs, aggregating errors across an eval experiment, or root-causing behavior from a trace file or viewer session.
Skill Claude CodeCodex
Run and use the NOOA trace viewer — the web UI + OTLP receiver for browsing agent traces and eval results. Use when starting the viewer, importing/exporting/deleting traces, querying the viewer's REST API, or wiring an agent run so traces show up at localhost:5001.
Skill Claude CodeCodex
Render and refine the system prompt of a nooa agent. Use when the user wants to inspect, debug, or improve an agent's prompt, context blocks, or tool documentation.
Skill Claude CodeCodex
Explore and debug agent execution traces. Use when the user asks to analyze a trace, debug an agent run, investigate errors in a trace, or when they paste a trace-explorer prompt from the viewer UI.
Skill Claude CodeCodex
Guide for adding a new benchmark or training environment to NeMo-Gym. Use when the user asks to add, create, or integrate a benchmark, evaluation, training environment, or resources server into NeMo-Gym. Also use when wrapping an existing 3rd-party benchmark library. Covers the full workflow: data preparation…
Skill Claude CodeCodex
Manage stacked branches and pull requests with the gh-stack GitHub CLI extension. Use when the user wants to create, push, rebase, sync, navigate, or view stacks of dependent PRs. Triggers on tasks involving stacked diffs, dependent pull requests, branch chains, or incremental code review workflows.
Skill Claude CodeCodex
Use when analyzing NeMo Gym benchmark rollouts for BLADE-style reports, writing benchmark methodology notes, checking whether a benchmark is BLADE-ready, comparing model runs, or explaining why a benchmark report passed, failed, or changed. Covers aggregate metrics, rollout evidence, report structure, root-cause…
Skill Claude CodeCodex
Use when debugging a Nemo Gym run or reward profiling job. Covers rollout collection failures, empty or partial JSONL outputs, stale materialized inputs, verifier/schema errors, Ray or Slurm issues, vLLM readiness, judge failures, tool/sandbox failures, cache problems, and throughput bottlenecks.
Skill Claude CodeCodex
Maintain the NeMo Gym Fern docs site — add, update, move, or remove pages under fern/. Use for any documentation change. Triggered by: "edit docs", "add doc page", "update docs", "rename page", "fix broken link", "add redirect", "preview docs", "publish docs", any request that touches fern/.
Skill Claude CodeCodex
Use when creating, validating, or documenting Nemo Gym pivot datasets from rollout, trajectory, chat-completion, Responses API, or tool-call artifacts. Covers Gym Responses-style row conversion, reconstructing model calls from flattened rollout output, parallel tool-call (functioncallbatch) labels, reasoning placement…
Skill Claude CodeCodex
Use to help users get started with Nemo Gym reward profiling. Covers the basic gym env start, gym eval run, and gym eval profile workflow, repeated rollouts, materialized inputs, rollout JSONL artifacts, task and rollout identity, output inspection, partial profiling, and rolloutinfos. For failed jobs, prefer…
Instructions file CodexOpenCode
Instructions for NVIDIA-NeMo/Gym, covering agents.md, quality bar, what this is, architecture and creating environments.
Instructions file
Instructions for NVIDIA-NeMo/Gym, a project described as: Evaluate and improve models and agents using environments.
Skill Claude CodeCodex
Simplicity gate for ALL code written, fixed, or reviewed in this repo. Invoke BEFORE writing any code change and when reviewing any diff. Treats over-complex or oversized code as a correctness bug, not a style issue.
Instructions file CodexOpenCode
Instructions for NVIDIA-NeMo/labs-molt, covering molt — working rules for ai assistants, code standards (hard rules, not preferences), comments and workflow.
Instructions file
Instructions for NVIDIA-NeMo/labs-molt, a project described as: An agentic-first RL framework for research (9k lines).
Hook Claude Code
Runs when you submit a prompt, before the agent sees it, running an inline shell check. From NVIDIA-NeMo/Megatron-Bridge.
Settings file Claude Code
Agent settings declaring 1 hook event (UserPromptSubmit).