Run OSINTNova investigations from your AI assistant
Connect Claude, ChatGPT, Cursor and other MCP-compatible assistants to your OSINTNova account. Ask in plain language and your assistant runs the right commands, waits for results and reports the evidence back to you.
- Claude
- ChatGPT
- Codex
- Claude Code
- Cursor
- VS Code
- Gemini CLI
Check whether analyst@example.com appears in any known breaches, then look for accounts registered to it.
How it works
The Model Context Protocol (MCP) is an open standard that lets AI assistants use tools provided by other services. OSINTNova is available as an MCP server, so a connected assistant can run the same commands you use on the platform.
-
01
Paste the setup prompt
Choose your assistant below and paste its prompt into your AI chat. It handles the setup it supports and guides you through any required settings.
-
02
Sign in and approve
Open the sign-in window from your assistant or its connector settings. Sign in with Google or Discord and approve the app by name.
-
03
Ask in plain language
Describe what you need. The assistant selects commands, follows each run to completion and presents the findings.
Available commands
20 investigation commands, run on OSINTNova under your account. Your assistant chooses the right command from your request.
- FreeNo credits used
- CreditUses a shared credit or Pro allowance
- ProPro members, or one credit on Free accounts
People and accounts
-
Username Intelligence
usernameDiscover candidate accounts and enrich available public profiles for one handle across supported platforms.
-
Email Intelligence
email-scanInvestigate available account presence, profiles and associated evidence for a full email address.
-
Breach Intelligence
email-pwnCheck known breach or infostealer exposure for a specific email address.
-
People Intelligence
peopleFind candidate public people records from a first and last name, optionally narrowed to a US state.
-
Phone Intelligence
phoneInspect a specific telephone number for available service and associated public evidence.
-
Discord Intelligence
discordInspect a known Discord account ID or resolve a public invite to its server context.
-
Steam Intelligence
steamInspect a known Steam account using its stable ID or vanity profile identifier.
-
Dark Web Intelligence
darkwebFind indexed mentions of an exact email, handle, domain or distinctive phrase in dark-web material.
Infrastructure and assets
-
Domain Intelligence
domainBuild a broad infrastructure and website report for a domain.
-
IP Intelligence
ipInvestigate an observed public network address for infrastructure, provider and approximate location context.
-
URL Detective
url-detectiveUnderstand where a public link leads and inspect the response observed at that URL.
-
Archive Intelligence
archiveFind historical captures of a particular public page or domain to compare evidence over time.
-
Email Header Analysis
headersExamine message routing and authentication for a supplied original email header block.
-
Vulnerability Intelligence
vulnerabilityResearch a CVE, CPE 2.3 identifier or specific vendor/product/version for known vulnerability context.
-
Who Am I
whoamiInspect the network footprint visible for this MCP connection without supplying a target.
-
Crypto Wallet
cryptoInspect a known wallet address or request general market context without a target.
Research and records
-
Google Dorks
google-dorksGenerate categorized search operators for an identified target without using AI query generation.
-
DorkGPT
dorkgptTurn a specific research goal into search queries when fixed Google Dorks categories are insufficient.
-
Image Intelligence
imageInspect an image the user supplies for metadata and available provenance clues, optionally requesting a cleaned copy.
-
Vehicle Intelligence
vinDecode a known vehicle VIN or resolve a US plate together with its issuing state.
Assistants can also call get_usage to check your remaining allowance, get_command_guide for full command documentation and get_operation to follow a running investigation. None of these use credits. Oracle, YouTube, Facebook and Instagram are available on the platform only.
Set up your assistant
Choose your assistant, copy its setup prompt and paste it into your AI chat. Local agents can configure the connection for you; some clients require a step in their connector settings. You complete sign-in and approval in your browser.
Paste this into a local Codex chat. With access to the Codex CLI on the same host, the agent can add the server and launch browser sign-in.
Connect this Codex host to OSINTNova at https://app.osintnova.com/mcp using Streamable HTTP and OAuth. Inspect the existing osintnova entry with codex mcp get osintnova. Reuse it if it already points to this URL; if absent, run codex mcp add osintnova --url https://app.osintnova.com/mcp. Resolve a conflicting entry with me instead of overwriting it. Let any sign-in started by add finish; otherwise run codex mcp login osintnova when authentication is needed. Run these commands yourself when the local CLI is available, and leave the process alive during browser approval. If the CLI is unavailable, use Codex's MCP settings or carefully merge [mcp_servers.osintnova] with url = "https://app.osintnova.com/mcp" into the active host's Codex config.toml, preserving all existing content. Do not assume a cloud shell configures my desktop. Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute. Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox. After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
Claude web, Desktop and Cowork use a connector added in Claude’s settings. A chat prompt can guide you through this required step. Access depends on your plan and workspace permissions.
- In Claude web, Desktop or Cowork, open Settings (or Customize) › Connectors.
- On an individual plan, select Add custom connector, name it OSINTNova and paste the server address. On Team or Enterprise, an owner first adds it in Organization settings › Connectors; members then select that OSINTNova connector.
- Select Connect, then sign in and approve Claude in the window that opens. Enable the connector in your chat; start a new session if its tools are not available.
Help me connect Claude web, Desktop or Cowork to OSINTNova at https://app.osintnova.com/mcp using its remote custom connector. For an individual plan, guide me to Settings or Customize > Connectors > Add custom connector, enter OSINTNova and this server URL, then Connect. On Team or Enterprise, an owner must first add it in Organization settings > Connectors; members then connect to that existing connector. If my plan or workspace permissions block setup, explain the specific limitation. This account connector requires Claude's settings flow; do not claim a chat prompt installs it. Do not edit claude_desktop_config.json or install Claude Code or a bridge inside a Cowork sandbox. After I approve the browser sign-in, help me enable the connector and start a new session if needed. Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute. Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox. After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
ChatGPT on the web requires adding the server as a custom plugin. Availability depends on your account and workspace permissions.
- Open ChatGPT Plugins, select +, then Add custom MCP server.
- Name it OSINTNova, paste the server address and choose OAuth. Review the warning and select Create as a plugin.
- Install the plugin, complete browser sign-in when prompted, then select OSINTNova with @ in a chat. If your workspace restricts custom plugins, ask its administrator to enable access.
Help me connect ChatGPT on the web to OSINTNova at https://app.osintnova.com/mcp. The supported setup is ChatGPT Plugins at https://chatgpt.com/plugins > + > Add custom MCP server. Help me enter the name OSINTNova, this server URL and OAuth authentication, review the warning, create the plugin, install it and complete browser sign-in. I must complete the plugin setup in ChatGPT; do not claim that writing a local Codex configuration or posting a URL installs a hosted ChatGPT plugin. If workspace permissions prevent setup, explain the exact restriction. Once installed, help me select OSINTNova with @ in a chat. Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute. Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox. After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
The agent can add the server and launch browser sign-in with the local Claude Code CLI. Older versions may require Authenticate from /mcp.
Connect this local Claude Code host to OSINTNova at https://app.osintnova.com/mcp using its native HTTP MCP transport and OAuth. Inspect claude mcp get osintnova and reuse a matching entry. If absent, run claude mcp add --transport http --scope user osintnova https://app.osintnova.com/mcp yourself. Preserve other servers; ask before replacing a conflicting entry. Check the installed CLI's help and, if supported, run claude mcp login osintnova in an interactive terminal to launch browser sign-in. If this version lacks that command, give me the short fallback: enter /mcp in Claude Code, select osintnova, then Authenticate. Do not run slash commands in a shell. A headless claude -p or Agent SDK run cannot complete this interactive login. Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute. Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox. After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
Set up manually
- Add the server from your terminal.
- Run
claude mcp login osintnovain an interactive terminal. If your version lacks this command, start Claude Code, enter/mcp, select osintnova and choose Authenticate.
Cursor’s agent can add the configuration and use Cursor CLI for browser sign-in when installed. Otherwise, select Connect in Cursor.
Connect this local Cursor installation to OSINTNova at https://app.osintnova.com/mcp using native remote HTTP MCP and OAuth. Inspect ~/.cursor/mcp.json and the current project's .cursor/mcp.json for an existing osintnova entry. Reuse a matching entry; ask before changing a conflicting one. If absent, merge "osintnova": {"url": "https://app.osintnova.com/mcp"} into mcpServers in ~/.cursor/mcp.json, preserving every other entry and taking a backup before editing. If Cursor CLI is already installed on this host and its help supports it, run agent mcp login osintnova to launch browser sign-in. Confirm this is Cursor's CLI before running it. Otherwise, give me the short fallback: Cursor Settings > Tools & MCP > osintnova > Connect. Restart Cursor if it has not loaded the edited configuration. Do not install a bridge or authenticate an unrelated CLI.
Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute.
Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox.
After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
Set up manually
- Open Cursor Settings › MCP and add a new server, or add this entry to
~/.cursor/mcp.json.
{
"mcpServers": {
"osintnova": { "url": "https://app.osintnova.com/mcp" }
}
}
- Run
agent mcp login osintnovaif Cursor CLI is installed, or select Connect next to OSINTNova in Cursor. Sign in when prompted; restart Cursor if the server has not appeared.
Copilot in agent mode can configure your workspace. VS Code may ask you to trust, start or authenticate the server.
Connect this VS Code workspace to OSINTNova at https://app.osintnova.com/mcp using Copilot agent mode and native HTTP MCP with OAuth. Inspect the workspace and active user profile's MCP configuration for an existing osintnova entry. Reuse a matching entry; ask before replacing a conflicting entry. If absent and the local VS Code CLI is available, run code --add-mcp with the JSON argument {"name":"osintnova","type":"http","url":"https://app.osintnova.com/mcp"}, quoted correctly for this shell. Otherwise merge "osintnova": {"type": "http", "url": "https://app.osintnova.com/mcp"} into servers in .vscode/mcp.json, preserving other servers, inputs and settings and taking a backup before editing. Use VS Code's MCP: List Servers controls to start osintnova and launch OAuth sign-in. Let me accept the server trust prompt. If those editor controls are unavailable to you, give me only the necessary Start or Authenticate step. Respect workspace and organization MCP policies and ensure OSINTNova tools are enabled in Copilot Chat.
Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute.
Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox.
After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
Set up manually
- Add this to
.vscode/mcp.jsonin your workspace, or run MCP: Add Server from the Command Palette.
{
"servers": {
"osintnova": { "type": "http", "url": "https://app.osintnova.com/mcp" }
}
}
- Start the server, sign in when prompted and use OSINTNova tools from Copilot Chat in agent mode.
This is for Gemini CLI on your computer. Its agent can add the server; browser sign-in uses the CLI’s /mcp auth command.
Connect this local Gemini CLI host to OSINTNova at https://app.osintnova.com/mcp using Streamable HTTP and OAuth. Inspect ~/.gemini/settings.json and the project's .gemini/settings.json for an existing osintnova entry. Reuse a matching entry; ask before replacing a conflicting one. If absent, run gemini mcp add --scope user --transport http osintnova https://app.osintnova.com/mcp using the installed CLI. If that command is unavailable, merge "osintnova": {"httpUrl": "https://app.osintnova.com/mcp"} into mcpServers in ~/.gemini/settings.json, preserving all other settings and taking a backup before editing. Enter /mcp reload in Gemini CLI to load the server, then /mcp auth osintnova to launch browser authorization. If you cannot operate interactive slash commands, tell me those required commands to enter. Do not run them in a shell or claim this configures the Gemini website. Browser OAuth needs access to the CLI host's localhost callback; a remote or headless container may require a local session.
Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute.
Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox.
After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
Set up manually
- Add this entry to
~/.gemini/settings.json.
{
"mcpServers": {
"osintnova": { "httpUrl": "https://app.osintnova.com/mcp" }
}
}
- In Gemini CLI, enter
/mcp reload, then/mcp auth osintnovato sign in.
Support depends on the client running your AI model. It needs remote Streamable HTTP and OAuth.
Connect the AI client running this chat to OSINTNova at https://app.osintnova.com/mcp. First identify this client's documented support for remote Streamable HTTP MCP and OAuth. If you can access its persistent host configuration, use the native setup mechanism and preserve existing settings. Otherwise guide me through the shortest supported connector-settings path. Do not assume the model's name proves client compatibility, and do not install another client's CLI or a bridge as a substitute. If this client lacks the required transport or OAuth support, say so clearly and recommend a supported client. Use the client's own OAuth flow. If it gives you an authorization URL, open that exact URL in my default browser or a new tab in my existing browser using available browser controls. Keep any login process and callback listener running until I finish. Let me sign in and approve access myself. Never invent an authorization URL or use the connection-management page as a substitute. Preserve my other servers, settings and approval controls. Do not ask me to paste passwords, API keys, access tokens or refresh tokens into chat. If this chat cannot access the persistent client configuration or complete its browser callback, explain the limitation and give me only the required manual step; do not configure an unrelated sandbox. After authorization, discover the OSINTNova tools and call get_usage and get_command_guide without running an investigation or spending credits. Only say the connection is ready after a real authenticated tool response. If a reload or new chat is needed to load tools, tell me to do that and repeat this verification there; configuration or login success alone is not proof.
- Use any MCP client that supports remote servers over Streamable HTTP with OAuth sign-in.
- Add the server address below. The client discovers sign-in automatically; no API key is needed.
Account and security
Uses your existing plan
Your assistant shares your Free credits or Pro allowances. There is no separate subscription, and free commands remain available when credits run out.
No passwords or API keys
You sign in on OSINTNova, never inside the chat. Each assistant is approved by name and can be revoked at any time.
Short-lived results
Completed results are stored encrypted and remain available to your assistant for one hour before they are removed.
Manage your connections
Sign in to see which assistants are connected to your account and revoke access. You can also connect first; your assistant will ask you to sign in.