Feedback1

Customer feedback and public roadmap MCP for B2B SaaS — query feedback, roadmap, changelog, and CRM context from Claude or Cursor.

View on feedback1.ai

Hosted MCP Server

npx add-mcp 'https://app.feedback1.ai/api/mcp'

Installs into Claude Code, Codex, Cursor and more

Documentation

What MCP is here

The Model Context Protocol (MCP) is a standard way for AI assistants to call tools and pull context from apps. Feedback1 runs a remote MCP server at https://app.feedback1.ai/api/mcp so supported clients can query your workspace instead of pasting CSV exports into chat.

The server is listed in the official MCP Registry as ai.feedback1/mcp (verified 26 Oct 2026 via the registry API; latest 1.1.0, with streamable-http and SSE remotes). Directories such as Glama or community awesome lists are listing targets, not launch sites. Do not treat a directory badge as a requirement to use Feedback1 MCP.

If you need plain HTTP for Zapier, scripts, or webhooks instead of an assistant, use the Public API. For a short educational primer on tools and resources, see what MCP tools and resources are.

Who it is for

Product managers who already plan in Claude or Cursor and need live customer evidence beside the PRD, not a stale spreadsheet.

Eng or AI-workflow owners wiring assistants to a feedback system of record so themes, features, and account context stay current.

Teams comparing vendors after seeing Canny or peers advertise MCP, and asking whether the feedback tool they use (or are evaluating) exposes an assistant endpoint.

This page is not for replacing Zendesk, Intercom, or Freshdesk. Keep those for tickets, chat, macros, and SLAs. Feedback1 can bridge support signal into product work (Feedback1 for Support), but MCP does not turn the assistant into a helpdesk. It is also not for unsupervised agents that invent roadmap decisions. Humans remain accountable for product approvals.

If your primary job is feedback collection into a shared workspace with a customer feedback portal, this integration is the assistant-side query layer on top of that loop. Browse other Feedback1 integrations for Jira, Azure DevOps, CRM, and messaging.

What you can do (verified tools)

As of the live MCP setup docs on 26 Oct 2026, assistants connected over MCP can:

  • Search feedback by keyword or semantic meaning
  • Find and list roadmap features
  • Look up CRM accounts and contacts
  • Get feedback statistics and top-voted features
  • Retrieve specific items by ID

Documented tool names on that page:

  • Discovery: list_products
  • Search: search_feedback, search_features, search_accounts, global_search
  • Retrieval: get_feedback, get_feature
  • Lists: list_feedback, list_features, list_accounts, list_contacts
  • Analytics: get_feedback_stats, get_top_voted_features
  • CRM: get_account, get_contact, and related feedback/feature lookups (as phrased on docs)

Re-check the docs page at publish for the authoritative list. Do not treat this marketing page as a second tool catalog if product and docs diverge; prefer updating docs first.

What MCP cannot do (buyer language):

  • Not a helpdesk: no ticket queues, live chat macros, or SLAs
  • Not autonomous product decisions or automatic PRD approval (feature-linked PRD drafts stay in the product UI with human sign-off)
  • Not a promise that every MCP client behaves identically; Feedback1 supports OAuth clients and documents an API-key fallback in docs
  • Not a fixed public "N tools" marketing number beyond what docs (or a dated public registry listing) state at publish

Screen recording: Claude/Cursor + Feedback1 MCP (60-90s) pulling a clustered theme or top-voted features over MCP - placeholder: coming with publish. Keep the on-page config block identical to the tested recording when the film ships.

Setup overview

Short path for OAuth-capable clients (Cursor and Claude):

  1. Open Claude or Cursor MCP settings.
  2. Add remote URL https://app.feedback1.ai/api/mcp.
  3. Sign in and pick the product to authorize.
  4. Ask for themes, top-voted features, or account-linked feedback.

Every detail (SSE endpoint, API keys, product-scoped header, troubleshooting, security, and rate limits) lives on the full MCP setup docs. Logged-in users can also copy ready-made configs from Settings → MCP in the app.

Copy-paste config

Tested starting config for OAuth-capable clients (same shape as docs; keep in sync at publish):

{
  "mcpServers": {
    "feedback1": {
      "url": "https://app.feedback1.ai/api/mcp"
    }
  }
}

For API-key / mcp-remote variants (Claude Desktop command-based configs, SSE URL, X-Feedback1-Product-ID), use the variants on the MCP setup docs rather than pasting three full blocks here.

How peers market MCP

Many feedback vendors now expose MCP. This note is honest and dated 26 Oct 2026. It is not a fake feature matrix and does not invent peer tool counts or pricing.

  • Canny markets MCP at canny.io/features/mcp plus a Help Center article, for ChatGPT, Claude, and Cursor, with OAuth and read/write within permissions (paid-plan availability per their FAQ language). If you are choosing a board vendor, not only an MCP endpoint, see Canny alternatives.
  • Quackback documents MCP for feedback triage, changelog, and (when modules are on) support inbox / help center, with cloud availability from Pro per their note (quackback.io/docs/mcp).
  • UserJot markets MCP on comparison and product copy (api.userjot.com/v1/mcp, API token; agents read/triage requests, roadmap, changelog, conversations per their live marketing page as of 26 Oct 2026). Their /docs/mcp URL returned 404 on that fetch; cite their live marketing page, dated.
  • VoteShip publishes install and developer docs (voteship.app/install.md, remote https://voteship.app/mcp, npm @voteship/mcp-server), with public tools without a key and admin tools with an API key per their docs.

Feedback1 is not the third-party packages interactive-feedback-mcp or mcp-feedback-enhanced. Those are separate projects that sometimes appear in adjacent search queries.

Choose Feedback1 when the system-of-record job is AI themes, CRM-weighted product context where configured, public roadmap and changelog, and flat workspace pricing with a self-serve trial. Skip Feedback1 MCP when you need a helpdesk-first suite.

Pricing

Verified on /pricing (confirm again at publish):

  • Free: $0 per month
  • Startup: $99 per month
  • Business: $499 per month
  • Enterprise: custom
  • Paid plans: 14-day free trial, no credit card required

Frequently asked questions

What is the Feedback1 MCP server?

A Model Context Protocol endpoint so supported AI assistants can call Feedback1 tools (search feedback, list features, CRM lookups, and related reads documented on our MCP docs) against your workspace.

Does Feedback1 work with Claude and Cursor?

Yes. Docs recommend OAuth with URL https://app.feedback1.ai/api/mcp for Cursor and Claude. Older clients can use the SSE and API-key path documented on the MCP setup docs (/docs/integrations/mcp).

Is MCP a replacement for the Feedback1 UI?

No. Use the product UI for day-to-day triage, roadmap edits, and publishing. Use MCP to bring the same data into assistant workflows.

Can MCP approve roadmap changes on its own?

No. MCP is a query and handoff layer for assistants. Humans remain accountable for product decisions and approvals.

Is Feedback1 a helpdesk?

No. Feedback1 connects customer signal to product work. Keep Zendesk, Intercom, or Freshdesk for tickets, chat, macros, and SLAs. More: /for-support.

How is this different from Canny MCP?

Canny markets its own MCP for ideas and feedback inside ChatGPT, Claude, and Cursor. Feedback1's wedge is the feedback-to-product loop (themes, CRM context where configured, roadmap, changelog, flat $99 / $499 workspace pricing, 14-day trial). See /alternatives/canny for the broader bake-off.

Where do I get setup steps and the full tool list?

The MCP setup docs at /docs/integrations/mcp. Re-check that page at publish for the authoritative tool names.

How much does Feedback1 cost?

Free plan at $0, Startup $99 per month, Business $499 per month, Enterprise custom, on Feedback1 pricing (/pricing). Paid plans have a 14-day trial, no card.

Ready to ground assistants in your feedback system of record?

Pull clustered themes, roadmap context, and CRM lookups into Claude or Cursor over MCP, then keep triage and approvals in Feedback1.

Start your 14-day free trial, no card required · Full MCP setup docs · See pricing