agentleFS
Sign inStart free
NVIDIAKnown publisherNVIDIA/physicsnemo-curator/docs/developer-guide/agents.md

physicsnemo-curator / developer-guide

AGENTS.md67 starsChanged 6 months agoApache-2.0View on GitHub

PhysicsNeMo Curator ships with Claude skills — structured instructions that guide an AI coding agent through implementing new components. If you use Claude Code (or a compatible agent), these skills automate much of the boilerplate and ensure new code follows project conventions from the start. Five skills live in .claude/skills/: 1. Open the repository in Claude Code (or any agent that supports Claude skills). 2. Describe what you want to build — for example: "Add a new source that reads…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,542 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository6 other files it gives its agents.

The file

We split this agents.md into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Test scaffolding with pytest.mark.requires

    line 51

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
agents.md
<!--- # SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. # SPDX-License-Identifier: Apache-2.0 -->

Using AI Agents

PhysicsNeMo Curator ships with Claude skills — structured instructions that guide an AI coding agent through implementing new components. If you use Claude Code (or a compatible agent), these skills automate much of the boilerplate and ensure new code follows project conventions from the start.

Available Skills

Five skills live in .claude/skills/:

SkillTriggerWhat it does
add-source"Add a new data source"Generates source class with params, tests, registry
add-filter"Add a new filter"Generates filter with correct generator patterns and tests
add-sink"Add a new sink"Generates sink with naming strategy, append logic, and tests
curator-reviewer"Review this PR"Reviews PRs against project standards (8-pass checklist)
testing"Run tests"Guides running Python/Rust tests with correct tooling

How to Use Them

  1. Open the repository in Claude Code (or any agent that supports Claude skills).

  2. Describe what you want to build — for example:

    "Add a new source that reads Parquet files from S3 and yields Mesh objects."

  3. The agent detects the matching skill and follows its guided workflow:
    • Asks discovery questions (file format, domain, naming, etc.)
    • Generates the implementation with correct patterns
    • Creates tests, registers the component, and runs quality checks

What the Skills Automate

Each skill encodes the project's conventions so you don't have to remember them all:

  • SPDX license headers
  • from __future__ import annotations and TYPE_CHECKING import patterns
  • ClassVar declarations for name and description
  • params() classmethod with proper Param objects
  • NumPy-style docstrings (99 % coverage requirement)
  • Generator semantics (sources and filters)
  • Test scaffolding with pytest.mark.requires
  • Registry registration in __init__.py
  • Conventional Commit messages

Without an Agent

The skills are plain Markdown files — you can read them directly for reference even if you're not using an AI agent:

cat .claude/skills/add-source/SKILL.md
cat .claude/skills/add-filter/SKILL.md
cat .claude/skills/add-sink/SKILL.md
cat .claude/skills/curator-reviewer/SKILL.md
cat .claude/skills/testing/SKILL.md

They serve as detailed checklists for implementing components by hand.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free