> ## Documentation Index
> Fetch the complete documentation index at: https://www.oplane.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Oplane agent plugins

> What the Oplane plugin adds to Claude, Cursor, Codex, GitHub Copilot CLI, and opencode, what it sends to Oplane, and how to get help when something doesn't work.

The Oplane plugin gives your AI assistant Oplane's threat modeling skills, a security agent, and a connection to the Oplane MCP server. This page describes what the plugin contains and what it sends to Oplane, and helps you fix common problems. To install the plugin, follow the steps for your tool in [Threat model from your IDE](https://www.oplane.com/docs/find/mcp.md#supported-environments).

## What the plugin contains

| Part | What it does | Available in |
| - | - | - |
| `analyze` skill | Threat models the current project, then checks each security requirement against your code and records the result in Oplane. | All supported tools |
| `analyze-pr` skill | Threat models the changes in a pull request and checks whether the change addresses each security requirement. | All supported tools |
| `security-analyzer` agent | A subagent your assistant can hand a full security review to. | Claude Code, Cowork, Cursor, GitHub Copilot CLI, and opencode. Codex has no subagents, so its plugin leaves the agent out. |
| Oplane MCP server | Connects your assistant to Oplane at `https://gravity.oplane.io/mcp/` so it can call the [MCP tools](https://www.oplane.com/docs/reference/mcp-tools.md). | All supported tools |

How you start a skill depends on your tool. In Claude Code you type `/oplane:analyze` or `/oplane:analyze-pr`, and in Cursor `/analyze` or `/analyze-pr`. Codex and opencode pick the skill from your request, so you can ask them to analyse the project or a pull request for security threats.

## Sign in

The Oplane MCP server signs you in with OAuth. When you connect the server in your tool, a browser window opens where you log in to Oplane, and tokens are then issued and refreshed for you. There are no tokens to copy.

To use the plugin from a CI pipeline or another tool where nobody signs in, create an [application](https://www.oplane.com/docs/administration/applications.md) and connect with its API key.

## What the plugin sends to Oplane

When you run an analysis, your assistant sends Oplane the title and description of the code or change it's threat modeling, which can include short code excerpts. It also sends git details, such as the repository URL, branch, commit SHA, and pull request URL, and the assessment it records for each security requirement, with file and line references.

The plugin sends nothing to any other service. It has no hooks or scripts that run on your machine, because the skills and the agent are instructions your assistant follows. Oplane stores the results in your account as described in the [privacy policy](https://www.oplane.com/legal/privacy-policy).

## Source code and license

The plugin is open source under the Apache License 2.0. You can read every skill and configuration file in the [oplane/oplane-plugin](https://github.com/oplane/oplane-plugin) repository on GitHub.

## Troubleshooting

### The assistant says the Oplane tools aren't available

The skills stop if they can't reach the Oplane MCP server, so check the connection in your tool and sign in again if it asks you to.

| Tool | Where to check |
| - | - |
| Claude Code | Run `/mcp`, select the Oplane server, and choose **Authenticate**. |
| claude.ai and Cowork | Open the Oplane plugin under **Customize > Plugins** and connect it on its **Connectors** tab. |
| Cursor | Open **Cursor Settings > MCP** and check that the Oplane server is enabled and connected. |
| Codex | Run `/plugins` to check the plugin, or `codex mcp login oplane` to sign in again. |
| GitHub Copilot CLI | Run `copilot plugin list` to check that the plugin is loaded. |
| opencode | Check the `oplane` entry in your `opencode.json`. See [Threat model from your IDE](https://www.oplane.com/docs/find/mcp.md#supported-environments). |

### The threat model isn't where you expected

Oplane adds threat models you create from your assistant to your personal workspace unless you name another workspace in your prompt. See [Choosing a workspace](https://www.oplane.com/docs/find/mcp.md#choosing-a-workspace).

## Get help

If something isn't working as expected and isn't covered above, email [support@oplane.io](mailto:support@oplane.io). Tell us which AI assistant you use, what you asked it to do, what you expected, and what happened instead.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.