Molt is a PyTorch-based reinforcement-learning framework for training software agents, including multimodal and multi-turn agents. It is intended for research using Ray for orchestration, vLLM for model rollouts, and NVIDIA AutoModel with FSDP2 for training.
Borrowing it
Nothing to install: this file belongs to NVIDIA-NeMo/labs-molt. 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/labs-molt/main/AGENTS.mdgit clone --depth 1 https://github.com/NVIDIA-NeMo/labs-moltWrote 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/instructions/nvidia-nemo/labs-molt/agents-md)<a href="https://agentmods.dev/instructions/nvidia-nemo/labs-molt/agents-md"><img src="https://agentmods.dev/badge/instructions/nvidia-nemo/labs-molt/agents-md/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/instructions/nvidia-nemo/labs-molt/agents-md"><img src="https://agentmods.dev/badge/instructions/nvidia-nemo/labs-molt/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00455 | $0.00455 |
| Opus 5.5 | $0.00182 | $0.00182 |
| Sonnet 5.5 | $0.00091 | $0.00091 |
| Haiku 4.5 | $0.00046 | $0.00046 |
Grade A, and why
labs-molt AGENTS.md 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 4d 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.
What it actually says
Molt — working rules for AI assistants
Code standards (hard rules, not preferences)
The priority order is explicit: human readability comes first; coding-agent traceability is the minimum gate. A human should understand code in one pass, and an agent must be able to trace a feature from CLI flag to executed branch, tensor/record, metric, and test without reconstructing hidden control flow.
- Over-complex or hard-to-follow code is a bug, not a style issue.
- Reduce complexity and line count — prefer deleting code over adding it.
- No over-encapsulation: a helper needs 3+ real call sites AND nontrivial logic; never wrap trivial code or add classes/files for one call site.
- Do NOT remove features, performance knobs, or observability in the name of simplicity — knobs default ON stay ON.
- A "bug" that cannot trigger under the shipped recipes is not worth fixing.
Details: .agents/skills/simplicity-first (invoke before any code change).
Comments
Concise "why" only, 2-4 lines, written for an external reader: no job ids, commit hashes, single-run metrics, or internal cluster paths; keep upstream issue/PR links.
Layout
Skills live once in .agents/skills/<name>/SKILL.md, the cross-tool location Codex, Cursor and
Gemini CLI read directly; .claude/skills (Claude Code) and .codex/skills (older Codex, Cursor)
are symlinks to that directory, and CLAUDE.md is a symlink to this file. A new skill is just its
directory under .agents/skills/.
Workflow
- Reviews report findings only; fixes ship as one minimal diff per issue after approval.
- Verify with
python -m compileall -q molt examples/python testsandpython -m pytest -q; shell scripts withbash -n. - Train–inference consistency work (bitwise rollout vs. trainer logprobs, mismatch
debugging, aligned kernels):
.agents/skills/train-inference-consistency.
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.
- 4d ago Changed · +7 lines · +94 tokens per session 89e6e9ab84ec
- 13d ago Changed · +2 lines · +37 tokens per session 14c83833f924
- 2mo ago First seen · 32 lines · 324 tokens per session scan A bd49158685de
labs-molt AGENTS.md is an instructions file published in the GitHub repository NVIDIA-NeMo/labs-molt (1,209 stars, last pushed 2d ago), licensed Apache-2.0. It adds 455 tokens to every session, about $0.0018 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.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.