Connect Signado to your AI assistant

Use this guide to connect your existing Signado workspace to Claude Code, Claude, ChatGPT, or Codex. The connection uses OAuth in your browser and reads the required client details automatically. You do not need an API key, OAuth client ID, or client secret.

Before you start

• You need a Signado account.
• You need access to the workspace you want the assistant to use.
• The remote endpoint is https://mcp.signado.io/mcp.

Claude Code

Run this command:

claude mcp add --transport http signado https://mcp.signado.io/mcp

Then authenticate:

1. In Claude Code, run /mcp.
2. Choose signado.
3. Click Authenticate.
4. Sign in with your Signado account in the browser.
5. Choose the workspace to connect.
6. Approve the connection.

Claude web app

1. Open Settings → Connectors → Add custom connector.
2. Name it Signado, enter https://mcp.signado.io/mcp as the server URL, and keep authentication on Always required.
3. Under OAuth client, keep Use Anthropic's hosted client metadata (Recommended).
4. Under Advanced, keep Streamable HTTP selected.
5. Select Add, then sign in with Signado, choose your workspace, and approve access.

ChatGPT

1. Open Settings → Plugins. If developer mode is off, enable it, then select + → New Plugin.
2. Name it Signado, enter https://mcp.signado.io/mcp as the server URL, and choose OAuth authentication.
3. Open Advanced OAuth settings and keep Client Identifier Metadata Document (CIMD) as the registration method.
4. Create the plugin, select Sign in with Signado, choose your workspace, and approve access.

Codex

Run this command:

codex mcp add signado --url https://mcp.signado.io/mcp
codex mcp login signado

The login command opens a browser approval flow. Sign in with your Signado account, choose the workspace, and approve the connection.

Codex 0.138.0 sends the OAuth resource parameter natively. If your Codex build cannot complete OAuth, add the resource fallback to ~/.codex/config.toml:

[mcp_servers.signado]
url = "https://mcp.signado.io/mcp"
oauth_resource = "https://mcp.signado.io/mcp"

Very old Codex builds may also need this top-level setting:

experimental_use_rmcp_client = true

Workspace selector

The browser approval flow includes a workspace selector. Choose the workspace you want Claude Code or Codex to use. MCP can only access that selected workspace, and the same app permissions apply to every tool call.

If you choose the wrong workspace, disconnect signado in the client and run the login flow again.

Verify the connection

After approval, the client should list 15 tools:

• ping
• list_warm_leads
• get_best_of_run
• get_warm_lead_detail
• export_leads
• preview_source
• add_source
• set_source_schedule
• remove_source
• find_email
• run_source_once
• cancel_source_run
• draft_message
• list_templates
• save_template

The deprecated compatibility aliases qualify and catchup_source remain callable for older clients but are intentionally absent from the tool listing. There are 17 callable tools in total.

Ask the assistant to call ping. A successful reply confirms that OAuth, workspace selection, and the remote endpoint are working.

Discovery sources

Your assistant can add, preview, and run four source types: keyword topics, creator or competitor profiles, Company pages, and individual LinkedIn posts. Use a personal linkedin.com/in/... URL for a creator and a linkedin.com/company/... URL for a Company page. Company pages remain a separate source type in Signado and appear as Company Pages in Warm Leads.

For example, ask: "Add https://www.linkedin.com/company/mondaydotcom/ as a Company page source." The assistant shows the plan and waits for confirmation before monitoring starts. You can include a Daily or Custom schedule when adding keyword, creator, or Company page sources, or change one later. For example: "Set my competitor scans to weekdays at 8am UK time." The assistant shows the exact schedule and waits for confirmation. Schedule edits change only the next future run; they never start a retroactive run. You can also preview a source first or run it once.

Where results appear

• Warm leads appear on /warm-leads.
• Found emails write back to the lead rows on /warm-leads.
• AI drafts from draft_message land in Campaigns at /campaigns.
• Discovery source changes appear in Settings under Discovery Sources.
• Credits show in each paid tool response as credits_charged, balance_after, and session_spend_total. Full usage history is in Settings -> Usage.

Nothing important lives only in the chat. If a session ends, saved results remain in the app.

Credit costs

The MCP connection uses the same credit pool and the same rates as the app:

Action	Cost
Read warm leads	0 credits
Export leads	0 credits
List or save templates	0 credits
Preview a source	1 credit
Add or remove a source	0 credits for the action
New warm lead found, enriched, and scored by an active source	5 credits
Optional one-lead re-score through the deprecated qualify alias	2 credits
Find an email	5 credits per contact
Draft an AI message	3 credits per contact

Company pages use the same profile-source pricing as creators: a preview or Daily scan costs 1 credit, and a successful scan refund follows the same rules. New warm leads cost 5 credits; re-engagement uses the rate configured for your plan. A hard failure before usable work is refunded, while a completed scan with no new posts or leads keeps only the work that was actually chargeable.