# Arcjet - MCP Server
[Arcjet](https://arcjet.com) is the runtime security platform that ships with your code. Enforce budgets, stop prompt injection, detect bots, and protect personal information with Arcjet's AI security building blocks.
The Arcjet MCP server gives AI coding assistants the skills of a security engineer. By connecting to the Arcjet MCP server, your AI coding tools can:
- **List teams** you belong to.
- **List sites** within a team.
- **Create new sites** within a team.
- **Get site keys** (`ARCJET_KEY`) for use in your projects.
- **List requests** received by a site with optional filtering.
- **Explain decisions** to understand why requests were allowed or denied.
- **Get request details** including headers, rules executed, and decision info.
- **Get site quota** usage and limits for the current billing window.
- **Analyze traffic** patterns, denial rates, top paths, and top IPs.
- **Detect anomalies** by comparing current traffic to the previous period.
- **Investigate IPs** with geo, threat intelligence, and request activity.
- **Get dry-run impact** analysis before promoting rules to live.
- **Get a security briefing** combining traffic, threats, anomalies, and recommendations.
- **List remote rules** configured for a site.
- **Create remote rules** with DRY_RUN or LIVE mode — configure rules with no code changes needed.
- **Update remote rules** by replacing the full rule configuration.
- **Delete remote rules** to immediately stop evaluation.
- **Promote remote rules** from DRY_RUN to LIVE after verification.
The MCP server is available at:
```
https://api.arcjet.com/mcp
```
It implements the latest
[MCP Authorization](https://modelcontextprotocol.io/specification/2025-06-18/basic/authorization)
and
[Streamable HTTP](https://modelcontextprotocol.io/specification/2025-06-18/basic/transports#streamable-http)
specifications with OAuth-based authentication.
## Supported clients
Any client that supports the latest [MCP
specification](https://modelcontextprotocol.io/specification/2025-06-18) with
[Streamable HTTP transport](https://modelcontextprotocol.io/specification/2025-06-18/basic/transports#streamable-http)
and [OAuth authorization](https://modelcontextprotocol.io/specification/2025-06-18/basic/authorization)
is supported. This includes:
- [ChatGPT](#chatgpt)
- [Claude Code](#claude-code)
- [Claude Desktop](#claude-desktop)
- [Cursor](#cursor)
- [VS Code with Copilot](#vs-code-with-copilot)
- [Windsurf](#windsurf)
- Many others that implement the MCP specification.
## Setup
### ChatGPT
1. In ChatGPT, go to **Settings**.
2. Navigate to **Connectors** and select **Add connection**.
3. Enter `https://api.arcjet.com/mcp` as the server URL.
4. Select **OAuth** for authentication.
5. Click **Create**.
ChatGPT handles the OAuth flow automatically.
### Claude Code
```bash
claude mcp add arcjet --transport http https://api.arcjet.com/mcp
```
Claude Code will open a browser for OAuth authentication on first connection.
Once authenticated, you can use the `/mcp` command to verify the connection.
### Claude Desktop
1. Open **Settings** in the sidebar.
2. Navigate to **Connectors** and select **Add custom connector**.
3. Configure the connector:
- **Name:** `Arcjet`
- **URL:** `https://api.arcjet.com/mcp`
### Cursor
Add to `.cursor/mcp.json` in your project:
```json
{
"mcpServers": {
"arcjet": {
"type": "streamable-http",
"url": "https://api.arcjet.com/mcp"
}
}
}
```
After adding the server, Cursor will show a **Needs login** prompt. Click it to
authorize Cursor to access your Arcjet account.
### VS Code with Copilot
Add to your `.vscode/mcp.json` in your project or user settings:
```json
{
"servers": {
"arcjet": {
"type": "http",
"url": "https://api.arcjet.com/mcp"
}
}
}
```
Or add it via the command palette:
1. Open the Command Palette (Ctrl+Shift+P on
Windows/Linux or Cmd+Shift+P on macOS).
2. Run **MCP: Add Server**.
3. Select **HTTP**.
4. Enter the URL: `https://api.arcjet.com/mcp`
5. Enter the name: `Arcjet`
6. Select **Workspace** or **User** depending on your preference.
VS Code will prompt you to authenticate via OAuth on first use.
### Windsurf
Add to your `mcp_config.json` file:
```json
{
"mcpServers": {
"arcjet": {
"serverUrl": "https://api.arcjet.com/mcp"
}
}
}
```
For more details, see the
[Windsurf MCP documentation](https://docs.windsurf.com/windsurf/cascade/mcp#adding-a-new-mcp-plugin).
## Available tools
Once connected, the following tools are available to your AI assistant:
- **`list-teams`** — Lists teams the authenticated user belongs to.
- **`list-sites`** — Lists sites within a specified team.
- **`create-site`** — Creates a new site within a specified team.
- **`get-site-key`** — Returns the SDK key (`ARCJET_KEY`) for a specific site.
- **`list-requests`** — Lists recent requests for a site. Supports filtering by conclusion (`ALLOW`, `DENY`, `ERROR`) and pagination.
- **`get-request-details`** — Returns full details for a specific request including headers, rules executed, and decision information.
- **`explain-decision`** — Explains why Arcjet allowed or denied a specific request. Returns a natural language summary, per-rule breakdown, and suggested next steps.
- **`get-site-quota`** — Returns quota usage and limits for a site in the current billing window.
- **`analyze-traffic`** — Analyzes request traffic over a time period. Returns total requests, denials, denial rate, top paths, top IPs, top denial reasons, and trend vs the previous period.
- **`get-anomalies`** — Detects unusual security patterns by comparing current traffic to the previous period. Identifies traffic spikes, geographic shifts, new threat activities, new bot signatures, risk escalation, and suspicious IP patterns.
- **`investigate-ip`** — Investigates an IP address in the context of a site. Returns geo location, threat intelligence (network type, threat activities, entity classification, risk level), and the IP's recent request activity (conclusion breakdown, denial reasons, targeted paths, daily timeline).
- **`get-dry-run-impact`** — Analyzes what would happen if dry-run rules were promoted to live. Shows how many currently-allowed requests would have been blocked by each rule type, which IPs would be most affected, and a false-positive estimate.
- **`get-security-briefing`** — Returns a comprehensive security briefing: active rules summary, traffic analysis, threat intelligence, anomaly detection, dry-run promotion readiness, quota status, and prioritized actionable recommendations. Designed for daily consumption.
- **`list-rules`** — Lists all remote rules configured for a site with their ID, type, mode, and configuration summary.
- **`create-rule`** — Creates a new remote rule for a site. Supports rate limit, bot, shield, and filter rule types.
- **`update-rule`** — Replaces an existing remote rule configuration. All fields must be provided (full replacement).
- **`delete-rule`** — Deletes a remote rule, immediately stopping it from being evaluated.
- **`promote-rule`** — Promotes a remote rule from DRY_RUN to LIVE mode after verification.
### Remote rules
Remote rules are managed through the MCP server or the Arcjet dashboard — no
code changes or redeployment needed. They apply globally to all requests for a
site. Only `rate_limit`, `bot`, `shield`, and `filter` rule types are supported
as remote rules. Rules that need parsed request body content (`email`,
`sensitive_info`, `prompt_injection`) require the SDK.
The most common use case for remote rules is responding to an active attack. For
example, if you notice suspicious traffic from a specific country, VPN, or IP
address, you can create a filter rule to block it immediately without deploying
new code:
1. Use `list-requests` to investigate the suspicious traffic and identify
patterns (e.g. a specific country, IP range, or VPN usage).
2. Use `create-rule` to add a filter rule in `DRY_RUN` mode to verify it
matches the right traffic. For example, block a specific country:
`ip.src.country == "XX"` (ISO 3166-1 alpha-2 country code e.g. `US`, `CN`,
`RU`), block VPN traffic: `ip.src.vpn`, or block an IP range:
`ip.src in { 1.2.3.0/24 }`.
3. Use `list-requests` again to confirm the rule is matching the expected
traffic without blocking legitimate users.
4. Use `promote-rule` to switch the rule from `DRY_RUN` to `LIVE`, immediately
blocking the attack traffic.
5. Once the attack subsides, use `delete-rule` to remove the block.
### Security monitoring
Use the analysis tools to maintain ongoing security awareness:
- **Daily briefing:** Call `get-security-briefing` periodically to get a
comprehensive overview of your site's security posture — traffic trends,
threat landscape, anomalies, dry-run readiness, quota status, and prioritized
recommendations — all in a single call.
- **Traffic analysis:** Use `analyze-traffic` to understand request patterns,
denial rates, top paths, and top IPs. This provides the same data as the
Arcjet dashboard analytics.
- **Anomaly detection:** Use `get-anomalies` to detect unusual patterns by
comparing current traffic to the previous period — traffic spikes, geographic
shifts, new threat activities, or suspicious IP behavior.
- **IP investigation:** When you identify a suspicious IP (from `analyze-traffic`
or `list-requests`), use `investigate-ip` to get full context: geo location,
threat intelligence, and the IP's complete request activity on your site.
- **Dry-run validation:** Before promoting a rule from `DRY_RUN` to `LIVE`, use
`get-dry-run-impact` to see exactly how many currently-allowed requests would
be blocked, which IPs are most affected, and a false-positive risk estimate.
## Usage examples
### Investigate and block suspicious traffic
> "I'm seeing a spike in denied requests on my site. Can you investigate what's
> happening and help me block the source?"
The assistant will call `analyze-traffic` to identify the spike, then
`list-requests` filtered to `DENY` to surface the top offending IPs. It uses
`investigate-ip` to pull threat intelligence for each IP, then
suggests a filter rule. Using `create-rule`, it creates the rule in `DRY_RUN`
mode so you can verify the match before calling `promote-rule` to go live.
### Get a daily security briefing
> "Give me a security briefing for my production site."
The assistant calls `list-teams` and `list-sites` to locate your production
site, then calls `get-security-briefing`. It returns a summary covering active
rules, traffic trends vs. the previous period, anomaly detection, threat
intelligence highlights, dry-run promotion readiness, quota status, and
prioritized recommendations you can act on immediately.
### Set up bot protection without redeploying
> "Add bot protection to my marketing site — start in dry-run mode so I can
> check it's not blocking real users."
The assistant calls `list-teams` and `list-sites` to find the site, then
`create-rule` to add a `bot` rule with `mode: DRY_RUN` configured to block
automated traffic. After traffic flows through, you can ask the assistant to call
`get-dry-run-impact` to see how many requests would have been blocked and
estimate false-positive risk, then `promote-rule` to switch the rule to `LIVE`
when you're satisfied.
## Authentication
The MCP server uses OAuth for authentication. When you first connect from any
supported client, you will be redirected to sign in with your Arcjet account.
Once authenticated, your AI assistant can securely access your account resources.
## Security
- **Verify the endpoint** — always confirm you are connecting to
`https://api.arcjet.com/mcp`.
- **Review tool calls** — enable confirmation prompts in your AI client so you
can review actions before they execute.
- **Trusted clients only** — only connect from AI clients you trust. Connecting
grants the AI tool the same access as your Arcjet account.
## Privacy and support
- **Privacy policy:** [arcjet.com/privacy](https://arcjet.com/privacy)
- **Support:** [support@arcjet.com](mailto:support@arcjet.com) or [join our
Discord](https://arcjet.com/discord). See our [support page](/support) for
more details.
