Add ModelOpt agent plugin marketplace

.agents/README.md

@@ -1,46 +1,49 @@
-# `.agents/` — agent-agnostic source of truth
+# `.agents/` — agent compatibility and shared config
 
-This directory is the canonical location for assets shared by AI coding agents
-working in this repository (Claude Code, Codex, Cursor, …).
+This directory exposes the ModelOpt plugin skills to repository-local agents
+and holds shared configuration.
 
 ## Layout
 
 ```text
 .agents/
-├── skills/                 # SKILL.md files (canonical)
-│   └── <skill-name>/SKILL.md
+├── skills → ../plugins/modelopt/skills
+├── plugins/
+│   └── marketplace.json   # Codex marketplace
 ├── scripts/                # shared helper scripts (sync-upstream-skills.sh, …)
 └── clusters.yaml.example   # remote-cluster config template
+
+plugins/modelopt/
+├── .claude-plugin/
+├── .codex-plugin/
+└── skills/                 # canonical SKILL.md files
+    ├── common/             # shared skill support files
+    └── <skill-name>/SKILL.md
 ```
 
 ## Why this exists
 
-Different agents look for skills/config in vendor-specific directories. Rather
-than maintaining N copies that drift out of sync, **`.agents/` is the single
-source of truth** — each agent's guidance or install mechanism points here
-directly.
+The installable plugin owns its skills, as required by the plugin format.
+Repository-local agent discovery uses relative symlinks to the same files.
 
 ## How each agent finds these
 
 Each agent points at `.agents/` through whatever mechanism it supports — never
 a copy:
 
 - **Claude Code** only auto-discovers skills under `.claude/skills/`, so
-  `.claude/` holds relative in-repo symlinks back into `.agents/`:
-  `.claude/skills → ../.agents/skills`, `.claude/scripts → ../.agents/scripts`,
-  and `.claude/clusters.yaml.example → ../.agents/clusters.yaml.example`. These
-  follow the same committed-symlink pattern already used elsewhere in this repo
-  (e.g. `CLAUDE.md`, `tools/launcher/modules/Model-Optimizer`).
-- **Future agents** (Codex, Cursor, …) add their own symlink or config pointing
-  at `.agents/`.
+  `.claude/skills/` holds relative symlinks into `.agents/skills/`.
+- **Repository agents** use `.agents/skills`, a relative symlink into the
+  plugin.
+- **Claude Code and Codex plugins** load `plugins/modelopt/skills` directly.
 
 ## Editing rules
 
-- **Always edit files under `.agents/`**.
+- **Always edit skills under `plugins/modelopt/skills/`**.
 - Vendored-verbatim skills (`launching-evals`, `accessing-mlflow`) are managed
   by `.agents/scripts/sync-upstream-skills.sh` — do not modify by hand.
-- New skills go in `.agents/skills/<skill-name>/SKILL.md` following the
-  conventions of existing skills (e.g. `.agents/skills/monitor/SKILL.md`).
+- New skills go in `plugins/modelopt/skills/<skill-name>/SKILL.md`.
+- Shared support files go in `plugins/modelopt/skills/common/`.
 
 ## Project-level cluster config
 

.agents/TOOLING.md

@@ -8,6 +8,11 @@ of the always-loaded agent instructions.
 Update `AGENTS.md` for repository-wide agent instructions. `CLAUDE.md` is
 symlinked to `AGENTS.md`, so changes there apply to both Codex and Claude Code.
 
+## Installable Skills
+
+The `modelopt` plugin packages the repository skills for use from any
+workspace. Installation commands are in the [README](../README.md#ai-agents).
+
 ## Local Overrides
 
 For private local instructions, use the tool-specific override file:

.agents/plugins/marketplace.json

@@ -0,0 +1,20 @@
+{
+  "name": "modelopt",
+  "interface": {
+    "displayName": "NVIDIA Model Optimizer"
+  },
+  "plugins": [
+    {
+      "name": "modelopt",
+      "source": {
+        "source": "local",
+        "path": "./plugins/modelopt"
+      },
+      "policy": {
+        "installation": "AVAILABLE",
+        "authentication": "ON_INSTALL"
+      },
+      "category": "Productivity"
+    }
+  ]
+}

.agents/scripts/sync-upstream-skills.sh

@@ -26,13 +26,14 @@
 #
 # Requires: gh, base64, awk. Run from the repo root.
 #
-# The script overwrites .agents/skills/<skill>/ with upstream contents and
+# The script overwrites plugins/modelopt/skills/<skill>/ through the
+# .agents/skills compatibility symlink and
 # re-applies our provenance lines into each SKILL.md frontmatter. If you have
 # local changes to a vendored skill, they will be lost — that is expected,
 # since vendored-verbatim skills should not be modified locally.
 #
 # Note: .claude/skills/ (and other agent-specific skill dirs) are symlinks to
-# .agents/skills/ — see .agents/README.md.
+# plugins/modelopt/skills/ — see .agents/README.md.
 
 set -euo pipefail
 

.agents/skills

@@ -0,0 +1 @@
+../plugins/modelopt/skills

.claude-plugin/marketplace.json

@@ -0,0 +1,21 @@
+{
+  "$schema": "https://json.schemastore.org/claude-code-marketplace.json",
+  "name": "modelopt",
+  "version": "0.1.0",
+  "description": "Model Optimizer agent plugins.",
+  "owner": {
+    "name": "NVIDIA Corporation"
+  },
+  "plugins": [
+    {
+      "name": "modelopt",
+      "source": "./plugins/modelopt",
+      "description": "Skills for Model Optimizer development, quantization, deployment, and evaluation.",
+      "version": "0.1.0",
+      "author": {
+        "name": "NVIDIA Corporation"
+      },
+      "category": "development"
+    }
+  ]
+}

.markdownlint-cli2.yaml

@@ -14,5 +14,5 @@ config:
 # Vendored upstream skills — kept byte-identical to upstream via
 # .agents/scripts/sync-upstream-skills.sh; do not reformat.
 ignores:
-  - ".agents/skills/launching-evals/**"
-  - ".agents/skills/accessing-mlflow/**"
+  - "plugins/modelopt/skills/launching-evals/**"
+  - "plugins/modelopt/skills/accessing-mlflow/**"

.pre-commit-config.yaml

@@ -70,10 +70,10 @@ repos:
         exclude: ^modelopt_recipes/configs/
 
       - id: sync-claude-skills
-        name: sync .claude/skills/ symlinks from .agents/skills/
+        name: sync .claude/skills/ symlinks from plugin skills
         entry: bash tools/precommit/sync_claude_skills.sh
         language: system
-        files: ^\.agents/skills/
+        files: ^plugins/modelopt/skills/
         pass_filenames: false
 
       - id: check-launcher-yaml

AGENTS.md

@@ -7,12 +7,10 @@ These instructions apply to AI-assisted work in this repository.
 - Start with `README.md` for project overview and install.
 - Use `modelopt/` for source, `tests/` for focused test coverage, and
   `examples/` or `docs/` for usage patterns.
-- **Agent skills and shared config live under `.agents/`** — the canonical,
-  agent-agnostic source of truth (`.agents/skills/<name>/SKILL.md`,
-  `.agents/scripts/`, `.agents/clusters.yaml.example`). Claude Code's
-  `.claude/skills`, `.claude/scripts`, and `.claude/clusters.yaml.example` are
-  relative symlinks into `.agents/`. Always edit files under `.agents/`, not the
-  symlink path. See `.agents/README.md` for the convention.
+- **Agent skills live under `plugins/modelopt/skills/`**, the installable
+  plugin's canonical skill tree. `.agents/skills` and `.claude/skills` expose
+  those skills through relative symlinks. Shared agent config and scripts
+  remain under `.agents/`. See `.agents/README.md` for the convention.
 
 ## Coding guidelines
 

README.md

@@ -170,7 +170,25 @@ Please read our [Contributing](./CONTRIBUTING.md) guidelines for details on how
 
 ## AI Agents
 
-For AI-assisted development setup, see the [agent tooling notes](./.agents/TOOLING.md).
+ModelOpt's agent skills can be installed from this repository and used in any
+workspace.
+
+### Claude Code
+
+```bash
+claude plugin marketplace add https://github.com/NVIDIA/Model-Optimizer.git
+claude plugin install modelopt@modelopt
+```
+
+### Codex
+
+```bash
+codex plugin marketplace add https://github.com/NVIDIA/Model-Optimizer.git
+```
+
+Then open `/plugins`, select the `modelopt` marketplace, and install `modelopt`.
+Contributors can also use the skills directly from a checkout. See the
+[agent tooling notes](./.agents/TOOLING.md).
 
 ### Top Contributors
 

plugins/modelopt/.claude-plugin/plugin.json

@@ -0,0 +1,20 @@
+{
+  "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
+  "name": "modelopt",
+  "displayName": "ModelOpt",
+  "version": "0.1.0",
+  "description": "Skills for Model Optimizer development, quantization, deployment, and evaluation.",
+  "author": {
+    "name": "NVIDIA Corporation"
+  },
+  "homepage": "https://github.com/NVIDIA/Model-Optimizer",
+  "repository": "https://github.com/NVIDIA/Model-Optimizer",
+  "license": "Apache-2.0",
+  "keywords": [
+    "modelopt",
+    "quantization",
+    "evaluation",
+    "deployment",
+    "llm"
+  ]
+}

plugins/modelopt/.codex-plugin/plugin.json

@@ -0,0 +1,37 @@
+{
+  "name": "modelopt",
+  "version": "0.1.0",
+  "description": "Skills for Model Optimizer development, quantization, deployment, and evaluation.",
+  "author": {
+    "name": "NVIDIA Corporation",
+    "url": "https://github.com/NVIDIA/Model-Optimizer"
+  },
+  "homepage": "https://github.com/NVIDIA/Model-Optimizer",
+  "repository": "https://github.com/NVIDIA/Model-Optimizer",
+  "license": "Apache-2.0",
+  "keywords": [
+    "modelopt",
+    "quantization",
+    "evaluation",
+    "deployment",
+    "llm"
+  ],
+  "skills": "./skills/",
+  "interface": {
+    "displayName": "ModelOpt",
+    "shortDescription": "Optimize, deploy, and evaluate models.",
+    "longDescription": "Provides Model Optimizer workflows for post-training quantization, deployment, evaluation, result comparison, and release validation.",
+    "developerName": "NVIDIA",
+    "category": "Developer Tools",
+    "capabilities": [
+      "Interactive",
+      "Write"
+    ],
+    "defaultPrompt": [
+      "Quantize this model with ModelOpt.",
+      "Deploy and evaluate this checkpoint.",
+      "Compare the baseline and quantized results."
+    ],
+    "brandColor": "#76B900"
+  }
+}

plugins/modelopt/skills/common/SKILL.md

@@ -0,0 +1,10 @@
+---
+name: common
+description: Shared ModelOpt support files. Use only when another ModelOpt skill directs you here.
+---
+
+# Shared ModelOpt Support
+
+This skill is a placeholder required for Codex plugin validation.
+
+Read only the file named by the calling ModelOpt skill.

pyproject.toml

@@ -211,7 +211,7 @@ extend-ignore = [
 "examples/*" = ["D"]
 "noxfile.py" = ["D", "E501"]
 "tests/*" = ["B017", "D", "E402", "PT012"]
-".agents/skills/*/tests/test_*.py" = ["D", "E402"] # Skill test scripts: docstring (D) + sys.path import-order (E402) exemptions
+"plugins/modelopt/skills/*/tests/test_*.py" = ["D", "E402"] # Skill test scripts: docstring (D) + sys.path import-order (E402) exemptions
 "*/_[a-zA-Z]*" = ["D"] # Private packages (_abc/*.py) or modules (_xyz.py)
 "*.ipynb" = ["D", "E501"] # Ignore missing docstrings or line length for Jupyter notebooks
 "modelopt/torch/kernels/*" = ["N803", "N806", "E731"] # triton style