Skip to main content

Harvestr MCP

How to connect and use the Harvestr MCP server

Written by Marina Salmon
Updated over 2 weeks ago

What is the Harvestr MCP Server?

The Harvestr MCP server lets AI assistants like Claude and Cursor directly query your Harvestr workspace — your feedback, discoveries, customers, and components — in natural language.

The server is read-only: your data is never modified.

Server URL

https://mcp.harvestr.io/mcp

Step 1: Generate an MCP token

  1. Go to Settings → Integrations → MCP Server

  2. Click New MCP Token

  3. Add a label to identify the client (e.g. "Claude Desktop", "Cursor")

  4. Click Create, then copy and store the token securely

Each token is scoped to your organization and can be revoked at any time from the same page.

Step 2: Connect your AI client

In your MCP compatible AI client (Cursor, Windsurf, etc.):

  1. Open your client's MCP settings

  2. Add a new server with:

  3. Save and reconnect

For Claude Desktop

On macOS

Open your Claude Desktop config file: ~/Library/Application Support/Claude/claude_desktop_config.json

Add the following (replace YOUR_TOKEN):

{
  "mcpServers": {
    "harvestr": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.harvestr.io/mcp",
        "--header",
        "Authorization: Bearer YOUR_TOKEN"
      ]
    }
  }
}

Save the file and restart Claude Desktop. You should see Harvestr appear in the tools panel.

On Windows

1. Open Command Prompt or PowerShell and install mcp-remote globally:

npm install -g mcp-remote

2. Find the full path to mcp-remote by running:

where mcp-remote

Copy the path ending in .cmd (e.g. C:\Users\YourName\AppData\Roaming\npm\mcp-remote.cmd).

3. Open your Claude Desktop config file: %APPDATA%\Claude\claude_desktop_config.json

Add the following (replace the path and YOUR_TOKEN with your own):

{
  "mcpServers": {
    "harvestr": {
      "command": "C:\\Users\\YourName\\AppData\\Roaming\\npm\\mcp-remote.cmd",
      "args": [
        "https://mcp.harvestr.io/mcp",
        "--header",
        "Authorization: Bearer YOUR_TOKEN"
      ]
    }
  }
}

Why the full path? Claude Desktop on Windows doesn't resolve commands from PATH the same way a terminal does. Using just mcp-remote will fail — you need the absolute path to mcp-remote.cmd.

Save the file and restart Claude Desktop. You should see Harvestr appear in the tools panel.

What you can do

The Harvestr MCP server gives AI tools access to the following data:

  • Feedback

  • Customers

  • Customer attributes

  • Discoveries

  • Components

All data is fetched live from your workspace at query time.

Below are the main things you can ask — organized by the type of work you're doing.

Explore your backlog

Browse and filter your Discoveries across components, states, and time periods.

Example prompts:

  • "List all open discoveries in the Billing component"

  • "Which discoveries are currently In Progress?"

  • "Show me everything we shipped in Q4"

  • "What discoveries were created in the last 30 days?"

  • "Do we have anything in our backlog about onboarding?"

Analyze feedback

Summarize, search, and make sense of the feedback linked to your discoveries — including both validated feedback and AI-suggested feedback pending review.

Example prompts:

  • "Summarize all feedback on the Dashboard performance discovery"

  • "What are customers saying about our mobile app?"

  • "Find feedback mentioning slow loading times"

  • "How many feedback items are linked to our SSO discovery?"

  • "Show me the most recent feedback we received on integrations"

Understand your customers

Look up which customers are behind a discovery, find everything a specific company has asked for, or segment by customer attributes.

Example prompts:

  • "Which customers have requested bulk export?"

  • "Summarize all feedback from Acme Corp"

  • "How many enterprise customers mentioned this issue?"

  • "Find all companies in the healthcare industry who submitted feedback"

  • "What are our top 5 customers by feedback volume asking for?"

Spot trends

Analyze feedback volume over time to identify spikes, drops, and emerging topics.

Example prompts:

  • "Has feedback on performance issues increased this quarter?"

  • "Which discoveries received the most new feedback in the last 30 days?"

  • "Show me feedback volume trends on our API discovery over the past 6 months"

  • "Are there any topics where feedback is growing fast right now?"

Prioritize and plan

Get data-driven input for prioritization decisions by combining feedback volume, customer data, and discovery status.

Example prompts:

  • "What are the top 10 most-requested open discoveries by feedback volume?"

  • "Which open discoveries have the most feedback but haven't been started yet?"

  • "What should I focus on next quarter based on what customers are asking for?"

  • "Give me a prioritized list of discoveries in the Analytics component"

A few tips for better results

  • Be specific about scope. Mentioning a component name, customer name, or time period helps the assistant narrow results faster and avoids it asking for clarification. "Feedback on the Integrations component in the last 60 days" will get you a better answer than "recent feedback".

  • Ask follow-up questions. The assistant maintains context within a conversation, so you can drill down: start with "What are my top discoveries by feedback volume?" then follow up with "Summarize the feedback on the top one".

  • Name your discoveries. The assistant searches discoveries by title, so the closer your discovery titles reflect the topic you're asking about, the more reliably it finds the right one. "Slow dashboard load times" will surface more easily than "Perf issue #3".

Available tools

These are the tools your AI assistant can call against your Harvestr workspace. Useful if you're building agents or agentic workflows on top of the Harvestr MCP.

Tool

What it does

list_harvestr_discoveries

Search and filter discoveries by title, component, state, or date. The primary tool for browsing your backlog.

list_harvestr_discoverystates

List all workflow states in your workspace (e.g. Backlog, In Progress, Done). Use to get state IDs for filtering discoveries.

list_harvestr_feedback

Search feedback by topic (semantic), customer, or discovery. Returns both validated and AI-suggested feedback.

list_harvestr_messages

Access full source messages (support tickets, Slack threads, notes). Use when you need the complete context behind a feedback extract.

list_harvestr_components

Browse your product component hierarchy. Use to scope discovery and feedback queries to a specific product area.

list_harvestr_customers

Look up users and companies by name. Use to find customer IDs before filtering feedback by requester.

search_harvestr_customers_by_attributes

Find customers by CRM attributes — plan tier, industry, geography, etc. Supports semantic search on text attributes and numeric filters.

analyze_harvestr_feedback

Analyze feedback volume over time. Returns a time series with period-over-period changes, globally or broken down by discovery.

A few things worth knowing when building workflows:

  • list_harvestr_feedback is the only tool with true semantic search — it finds conceptually related feedback even with different wording. All other tools use keyword/title matching.

  • Tools chain naturally: use list_harvestr_discoveries to get a discovery ID, then list_harvestr_feedback with discovery_ids to get its feedback, then list_harvestr_customers to profile who's behind it.

  • All tools support pagination via a cursor parameter for workflows that need to process large result sets.

  • The server is read-only — no tool can create, update, or delete data.

Security

  • Treat your MCP token like a password — never commit it to version control

  • Each token is scoped to your organization only

  • The server is read-only: no data can be created, edited, or deleted via MCP

  • Create one token per client so you can revoke individual access without disrupting other tools

  • Revoke tokens anytime at Settings → Integrations → MCP Server

Troubleshooting

"No tools available" after connecting Check that your token is correctly formatted with the Bearer prefix in the Authorization header.

Authentication errors Your token may have been revoked. Generate a new one at Settings → Integrations → MCP Server.

Slow responses on large queries Queries that span your full workspace (no date range or filter) may take longer. Scope your question to a time period or specific component for faster results.

Did this answer your question?