Peec AI

Query brand visibility across AI search engines and edit a Peec project.

Add to Agentman Studio

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

Resources

What the Peec AI connector does

The Peec AI MCP server connects a Peec AI workspace — an analytics product that tracks how brands appear in AI search engines — to Claude, ChatGPT and any MCP-compatible agent. Peec documents 76 tools that query visibility reports and edit project configuration. Sign-in is OAuth 2.0, or a personal access token for clients that cannot do OAuth.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

Peec AI tools (76)

  • list_projects
  • list_brands
  • list_prompts
  • list_topics
  • list_tags
  • get_brand_report
  • get_domain_report
  • get_url_report

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Anthropic's directory understates the surface by 68 tools. It lists 8 read tools and records permissions as Read; Peec documents 76 including 43 writes and 13 deletes. Nothing in the directory is invented — the gap is entirely under-listing — but the listing is not a boundary.
  • One opaque OAuth scope. Both the RFC 9728 resource descriptor and the authorization server metadata advertise scopes_supported as ["default"], checked live on 2026-08-22. No grant separates a report read from a cascading delete; the boundary that exists is Peec's own organization-owner check, applied server-side after consent.
  • We could not read tool schemas or annotations. The endpoint returned 401 to an anonymous request, so the parameter detail on this page is Peec's documentation, and the readOnlyHint and destructiveHint values are Peec's claims rather than our observation.
  • Some deletes are permanent and some cascade. delete_domain_classification and delete_url_classification are documented as permanent. delete_prompt soft-deletes the prompt's chats with it. delete_topic deletes AI-generated prompt suggestions under the topic while merely detaching the prompts.
  • Writes are metered and plan-gated. create_prompt may consume plan credits, unarchive_prompt is subject to the project's active-prompt limit, and a partially accepted batch returns a plan-credit warning string rather than failing outright.
  • Rate limited to 1,000 tool calls per minute per user, per Peec's tools reference. It states the server tells the assistant to retry after a few seconds on exceeding it.
  • Brand edits block on a background recalculation. Peec documents that changing a brand's name, regex or aliases triggers a metric recalculation, and that further edits during the recalc fail until it completes — minutes, not seconds.
  • Agent Analytics needs a separate data source. list_bots and get_agent_visits require a connected log feed via Cloudflare Workers or a CSV upload; without one they have nothing to aggregate.
  • Scraped content can lag or be absent. get_url_content returns null content where Peec has tracked a URL but scraping has not finished, which Peec says can take up to 24 hours, and 404s for URLs no Peec project has ever retrieved.
  • The changelog covers the REST API, not the MCP server. Peec's published changelog documents the Customer API — still labelled v0 beta, with breaking changes possible — and carries no MCP entries, so there is no dated record of when tools were added to this server.

Frequently asked questions

Is the Peec AI connector read-only?

No. Anthropic's directory labels its permissions Read and lists eight read tools, but Peec's own tools reference documents 33 read tools and 43 write tools, including 13 that delete. The directory listing is a stale subset. Assume the connector can edit your project configuration once you authorize it.

How many tools does the Peec AI MCP server have?

Peec documents 76: 33 read and 43 write. Anthropic's directory lists 8, all read. Every directory name appears in Peec's reference, so the directory is a strict subset rather than a conflicting list. We could not run a handshake to confirm the live count, because the endpoint requires authentication.

What OAuth scopes does the Peec AI MCP server request?

One, called default. The server's RFC 9728 descriptor and its authorization server metadata both advertise scopes_supported as a single default entry, checked live on 2026-08-22. There is no scope that grants reading a report without also granting the delete tools, so consent is all-or-nothing.

Can the Peec AI connector delete my tracked prompts?

Yes, if your account is an organization owner on the project. Peec documents delete_prompt as destructive and states it cascades to the prompt's chats, which are soft-deleted too. A bulk variant removes up to 50 at once. Peec flags every delete tool destructiveHint true so clients prompt first.

Which deletions in Peec AI are permanent rather than soft?

The custom classification ones. Peec documents delete_domain_classification and delete_url_classification as permanent deletes that cascade through the override table, so affected domains and URLs fall back to their heuristic classification. Brands, prompts, tags and topics are documented as soft deletes instead.

Do I need a paid Peec AI plan to use the MCP server?

You need an account with at least one project, and some writes are metered. Peec lists a Peec AI account with at least one project and a supported AI tool as the requirements. It also states that create_prompt may consume plan credits and that unarchive_prompt is subject to the project's active-prompt plan limit.

Can the Peec AI MCP server send email or post content?

No tool it documents sends anything. Every one of the 76 reads or edits Peec project data. Peec's own use-case docs do describe drafting outreach from cited URLs and adding a Gmail MCP server to the same session to send it, so the outbound step comes from a second connector, not this one.

What untrusted text does the Peec AI connector put in my agent's context?

Third-party web pages and AI-engine responses. get_url_content returns the scraped markdown of any URL Peec has indexed as an AI source, and get_chat returns full engine responses including ad placements with clickout URLs. Treat all of it as data, never as instructions to the agent.

Why do the Peec AI prompts not show up in my client?

Try removing and re-adding the integration. Peec documents seven slash-command prompts and states that if prompts don't appear after connecting, removing and re-adding the Peec AI integration refreshes the tool and prompt list. Anthropic's directory advertises no prompts for this connector, so the client list is the authority.

Sources

More Data & Research connectors

  • Airfield Directory

    Look up airfields, landing fees, fuel prices, pilot reports and METAR/TAF from an AI agent.

  • Airtable

    Read, write and restructure Airtable bases from your AI assistant.

Browse all MCP connectorsClaude and ChatGPT setup guideBrowse AI skills

Product names and logos belong to their owners. Agentman is not affiliated with a connector's maker unless the page says the connector is built and run by Agentman.

Use it in an agent

Put Peec AI to work.

Connect Peec AI once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Open in Agentman Studio