shruwd

Track how a brand shows up in AI answers, and act on the findings, through Shruwd's MCP server.

Shruwd measures a brand's AI-search visibility: how often AI answers mention it, its share of voice against competitors, and which AI crawlers read its site. This skill connects Shruwd's MCP server (source: https://github.com/ilpr/shruwd-mcp, MIT) so you can set up a brand, track prompts and competitors, and read results.

External service. Every tool call goes to shruwd.io and acts as the signed-in user, or as the holder of the API key. The user needs a Shruwd account. Access is bound to one workspace and carries the user's role in it.

Setup

Hosted, with sign-in (recommended)

The server runs at https://shruwd.io/mcp (Streamable HTTP), so there's nothing to install and no key to store:

bash

openclaw mcp add shruwd --url https://shruwd.io/mcp --transport streamable-http --auth oauth --oauth-scope "read write"
openclaw mcp login shruwd
openclaw mcp doctor shruwd --probe

login opens a browser. The user signs in to Shruwd and picks the workspace. Use --oauth-scope read for read-only access. The user can revoke access in Shruwd under Account → Connected apps.

Hosted, with an API key (mcporter)

bash

mcporter call 'https://shruwd.io/mcp.shruwd_get_workspace' --header 'Authorization=Bearer ${SHRUWD_API_KEY}' --name shruwd

Keep the single quotes so that mcporter reads SHRUWD_API_KEY from the environment, not the shell.

Local (stdio), pinned

bash

openclaw mcp add shruwd --command npx --arg -y --arg shruwd-mcp@0.1.9 --env SHRUWD_API_KEY="$SHRUWD_API_KEY"
openclaw mcp doctor shruwd --probe

Using it

1. Start with shruwd_get_workspace to see the plan, usage and existing brands.
2. New brand: shruwd_create_brand (needs owner), then shruwd_add_prompts and shruwd_add_competitor.
3. shruwd_run_measurement starts a cycle. Results arrive over the following hours, so don't report numbers until shruwd_list_cycles shows the cycle has closed.
4. Read the results with shruwd_get_visibility, shruwd_get_visibility_series and shruwd_get_crawlers. Call shruwd_get_visibility once per engine the brand measures and report each separately. Rates and share of voice come with a 95% interval and n; insufficient_data means too few answers to give a number, not zero.
5. Visibility also gives the brand's mention rate per question type (buying questions apart from how-to ones) and, for each tracked company, how often it's the first of them an answer names.

Nothing is hard-deleted. Archiving a brand stops scheduling and keeps its history, removing a prompt deactivates it, and removing a competitor closes it as of now.

Full docs: https://shruwd.io/docs/api/mcp-server · Machine-readable: https://shruwd.io/llms.txt