Connect MCP clients

Before you connect

1. Crawl a site in the Silkra desktop app.
2. Sync that workspace to the cloud (paid plans).
3. Keep the MCP URL handy:

https://mcp.usesilkra.com/api/mcp

Silkra MCP is read-only. Clients can list synced sites, filter pages, read content, and search — they cannot crawl or change your local project.

Sign-in happens on usesilkra.com. Approve access once per client; you may need to click Connect again in the client after the browser returns.

Claude

1. Open Settings → Connectors → Add custom connector.
2. Name it Silkra.
3. Paste the MCP URL above.
4. Leave OAuth Client ID and Secret blank.
5. Connect, sign in to Silkra, and allow access.

If Claude shows a brief connect issue after consent, click Connect once more — that retry usually finishes the handshake.

Cursor

Fastest path: paste this into a Cursor Agent chat and let it configure MCP for you:

Install the Silkra remote MCP server in Cursor.

1. Add (or merge) this into my user MCP config at ~/.cursor/mcp.json — create the file if it does not exist. If I already have mcpServers, keep my existing servers and only add/update "silkra":

{
  "mcpServers": {
    "silkra": {
      "url": "https://mcp.usesilkra.com/api/mcp"
    }
  }
}

2. Confirm the JSON is valid.
3. Tell me to reload MCP / restart Cursor if needed, then complete Silkra OAuth in the browser when Cursor prompts me (leave Client ID blank).

Manual option — add the same block yourself in ~/.cursor/mcp.json (global) or .cursor/mcp.json (project), or use Customize → MCP and paste https://mcp.usesilkra.com/api/mcp.

Codex

Fastest path: paste this into a Codex chat and let it configure MCP for you:

Install the Silkra remote MCP server in Codex.

1. Add (or merge) this into ~/.codex/config.toml — create the file if needed. Keep my other settings; only add/update the silkra MCP server:

[mcp_servers.silkra]
url = "https://mcp.usesilkra.com/api/mcp"

2. Run: codex mcp login silkra
3. I will approve access in the Silkra browser window. Confirm when the server is connected and list available Silkra tools.

Manual option — edit ~/.codex/config.toml (or project .codex/config.toml), then run:

codex mcp login silkra

Or:

codex mcp add silkra --url https://mcp.usesilkra.com/api/mcp

ChatGPT

1. Enable Developer Mode (Settings → Connectors → Advanced).
2. Add custom connector and paste the MCP URL.
3. Use OAuth (no API key).
4. Sign in to Silkra and allow access.

Custom connectors require a ChatGPT plan that includes Developer Mode. Directory listing is separate and not required for your own connector.

Starter prompts

Paste these after the client is connected. Replace the site name or topic as needed.

Inventory

Using Silkra MCP, list my synced sites and summarize page counts for each.

Issue triage

Using Silkra, what are the highest-severity SEO issues on [site]? Group them by theme and suggest a fix order.

Content gaps

Search my synced Silkra crawl for pages about [topic]. Which URLs are strongest, which are thin, and what titles would you improve?

Cannibalization

Find pages on my Silkra crawl that compete for the same intent around [query]. List overlapping URLs and a consolidation recommendation.

Brief from evidence

Pull page content for the top matching Silkra pages about [topic], then draft a short content brief with target URL, angle, and internal links to reuse.

What tools expose

Only the latest complete synced crawl run is published per site. Re-sync from desktop after a crawl to refresh what clients see.

Troubleshooting

Agents guide → (in-app agent vs remote MCP clients)