Skip to main content
Reechee runs a hosted MCP server so any AI client - Claude, Cursor, Codex, ChatGPT - can pull software-review pain points and opportunities straight into a conversation or an agent.
It speaks Streamable HTTP, so there’s nothing to install - point your client at the URL. It’s a thin wrapper over the REST API, so the data matches exactly. Every MCP call is made as a Reechee account: log in from your client (Connected) or send an API key. Start free, no card needed - pay-as-you-go includes the full toolset.

Connect your client

Add the server to Claude, Cursor, or Codex - takes about a minute.

​
Connected - log in with your account

Clients that support OAuth (Claude, ChatGPT, Cursor) log you in with your Reechee account instead of a key. Point them at https://mcp.reechee.io/mcp and the client opens the Reechee login by itself. The older https://mcp.reechee.io/mcp/auth URL behaves exactly the same, so existing setups keep working.
New to Reechee? Create your account and finish setup first, then connect. Signing up in the middle of Connect leaves your client waiting, and you will need to click Connect again.
The client opens a browser, you log in and approve a short “Connect to Reechee” screen, and it’s connected - nothing to paste. You get the full read + action set and the credit tools (all sixteen tools). The consent screen has an optional credit budget; leave it empty for no limit, or set one to cap what the connection can spend in any 30 days. A team still on a plan from before pay-as-you-go pricing gets three read tools (search_products, a pain-points taste, and get_competitors) until it moves to pay-as-you-go. See Connect Claude and the authentication guide.

​
With an API key

Send a key as Authorization: Bearer rch_live_… and the same server exposes sixteen tools. get_pain_points returns the full report (every theme, quote, and trend), and you get reads, safe write actions, and credit-spending tools: The point is chaining: “find new opportunities in my watchlist, unlock the best 3 buyers’ emails, and draft outreach,” or “follow the top 5 competitors of Salesforce.” The MCP tools accept page_size from 1 to 50 (the REST API allows up to 100). list_opportunities and get_opportunity return both review-sourced opportunities (software reviews on G2, Capterra, Trustpilot, TrustRadius, SourceForge) and social-sourced ones (LinkedIn and Reddit mentions) - each row’s source field says which. Social sources require a Scale plan or higher. For review-sourced opportunities, get_opportunity includes review_title, pros, cons, and general_comments under review. On pricing version 2 these are all null until your team reveals the reviewer, even if the company is already known. After reveal_reviewer, its nested opportunity.review and later get_opportunity calls return the full stored fields. Both the tool’s structured response and its human-readable text include the authorized evidence without truncation, so clients that read only text still receive it. Missing fields stay null in structured output and do not become invented text. Repeated reviewer reveals return the evidence without another debit. Legacy pricing version 1 retains access to stored review text before reveal. Organization and source access rules still apply, and list_opportunities does not include these full-review fields. See the REST evidence contract.
The four Credit tools spend Reechee credits. API keys always run behind a spend budget (a safety guardrail against an agent burning your balance): set a spend cap when you create the key under Settings → API keys. It cannot be added later, and a key with no cap gets 403 spend_cap_required. OAuth Connect: the consent screen has an optional credit budget. Leave it empty for no limit, or set a number to cap what the connection can spend in any 30 days.
A call without a login or a key is answered 401 with an OAuth discovery challenge, which is what makes OAuth clients open the login. See what pay-as-you-go includes and the authentication guide to create a key.

​
Next

Claude

Cursor

Codex