Skip to content

README.md

Latest commit

 

History

History
210 lines (136 loc) · 7.93 KB

README.md

File metadata and controls

210 lines (136 loc) · 7.93 KB
Raw

Roboflow Agent Plugin

Agent-ready Roboflow plugin that contains skills and MCP configuration for computer vision workflows: data management, training, evaluation, inference, model selection, Workflows, Universe, plans, and Roboflow platform APIs.

This repository is a plugin-shaped source of truth for AI agents (Claude Code, Codex, Cursor, OpenCode, and others). The canonical skill content lives in skills/; plugin manifests point at those files instead of copying them elsewhere.

Install as a plugin

The repo ships both plugin manifests pointing at the same skill content and MCP config:

Both manifests load skills from skills/ and bundle the Roboflow MCP server config from .mcp.json.

Claude Code

Install from GitHub — no clone required:

claude plugin marketplace add roboflow/computer-vision-skills
claude plugin install roboflow

The first command registers this repo as a marketplace source (run once per machine). The second installs the plugin.

Per-project installation

For per-project isolation — for example, when different projects need different ROBOFLOW_API_KEY values for different workspaces:

claude plugin install roboflow --scope local

Local scope writes the plugin into the current project only and reads the API key from that project's environment.

Install from a local clone 
git clone https://github.com/roboflow/computer-vision-skills
claude plugin marketplace add ./computer-vision-skills
claude plugin install roboflow

For a throwaway test without touching the installed-plugins list:

cd computer-vision-skills
claude --plugin-dir .

Codex

Add this repo as a marketplace source, then install the plugin from that marketplace.

Install from GitHub:

codex plugin marketplace add roboflow/computer-vision-skills
codex plugin add roboflow@roboflow

You can also install from the plugin browser after adding the marketplace:

codex /plugins

Choose the Roboflow marketplace source, select the Roboflow plugin, install it, and press Space if it is installed but still disabled. Restart Codex after installing or upgrading a plugin so cached skill metadata is refreshed.

Local clone workflow

When editing a local clone, register it as a local marketplace source:

git clone https://github.com/roboflow/computer-vision-skills
cd computer-vision-skills
codex plugin marketplace add .
codex plugin add roboflow@roboflow

Restart Codex after edits. If Codex still shows stale metadata, upgrade the marketplace or remove and re-add it:

codex plugin marketplace upgrade roboflow
codex plugin marketplace remove roboflow
codex plugin marketplace add .
codex plugin add roboflow@roboflow

If you registered the GitHub marketplace source instead, refresh it with codex plugin marketplace upgrade roboflow.

What the Codex marketplace file does

Codex reads .agents/plugins/marketplace.json, which points source.path at the repo root via ./. Codex resolves source.path relative to the marketplace root, so the plugin manifest in .codex-plugin/plugin.json, the skills in skills/, and the Roboflow MCP server config in .mcp.json are all loaded from this repository.

Codex caches installed plugins under ~/.codex/plugins/cache/, so a running Codex session may not see edits until Codex is restarted or the plugin is reinstalled from the Plugin Directory.

Codex CLI picks up ROBOFLOW_API_KEY from the shell environment that launches the codex binary. In Codex desktop, set the key in the local environment used by the workspace. Use a project-scoped .env if you need different keys per project.

One-line installer

The installer script registers this marketplace and installs the plugin for whichever supported agent CLIs it finds (codex, claude, or both). When using the published endpoint:

curl -fsSL https://repo.roboflow.com/agent-install/agent.sh | bash

PowerShell:

iwr -useb https://repo.roboflow.com/agent-install/agent.ps1 | iex

Install standalone skills

If you want the skills without the MCP server bundle — for example, with an agent that doesn't speak the plugin manifest format — install them directly:

npx skills add roboflow/computer-vision-skills

Install a single skill:

npx skills add roboflow/computer-vision-skills --skill inference

By default this installs into ./.claude/skills/ for the current project. Pass -g for ~/.claude/skills/ (global).

The npx skills CLI works with any agent that reads SKILL.md files from .claude/skills/ — Claude Code, Cursor, OpenCode, and others. See vercel-labs/skills for the full CLI reference.

Available skills

  • api-reference: REST API and inference API references
  • cloud-storage: connecting S3/GCS buckets to mirror images into a workspace
  • data-management: uploading images, labeling, dataset organization
  • inference: running inference, workflows, workflow templates
  • plans-and-pricing: Roboflow plans and credit usage
  • product-navigation: where features live in the Roboflow product
  • training-and-evaluation: training models and improving accuracy
  • universe: searching and using Roboflow Universe

Try it

After installing, start with one of these prompts:

  • "Build a Roboflow project for detecting defects in my manufacturing images."
  • "Run my saved Roboflow Workflow on a test image and save visual outputs to disk."
  • "Review my dataset and recommend the next model-training iteration."

MCP and skills

The Roboflow MCP server exposes live tools for projects, images, annotations, versions, models, Workflows, Universe, and feedback. Skills own the expert guidance and workflow playbooks.

That separation keeps the install model simple:

  • MCP server: live Roboflow tools and authenticated API access
  • Plugin skills: durable product guidance and workflow playbooks
  • This repo: canonical source for skill updates and plugin distribution

Limitations

  • Live MCP tools require ROBOFLOW_API_KEY in the environment used by your agent.
  • Product routes, model availability, credit rates, and pricing can change; verify current commercial details at roboflow.com/pricing, roboflow.com/credits, or the in-app billing pages.
  • Local skills are durable guidance, not a substitute for live MCP schema checks when a workflow block, API field, or tool parameter is uncertain.
  • The installer scripts require the target agent CLI (codex or claude) to already be installed.
Get your API key

Grab your Roboflow API key from the Roboflow settings: app.roboflow.com/settings/api

The key authenticates the bundled MCP server against https://mcp.roboflow.com via the x-api-key header.

Export it in the shell that launches your agent:

export ROBOFLOW_API_KEY=your_key

For persistence, add the export to your shell profile (~/.zshrc, ~/.bashrc) or to a project-local .env file loaded by your agent's environment. Per-project isolation is the safer default — keeps separate workspaces and billing accounts from leaking across projects.

Contributing

Skills are markdown. Open a PR with edits or a new folder under skills/. Each new skill must have a SKILL.md at its root with name and description frontmatter.

Run the structural validator before opening a PR:

python3 scripts/validate_plugin.py

License

Apache-2.0