Content
# MentionLeads MCP Server
Bring high-intent Reddit conversations, reply drafts, and real product feedback into your AI coding agent.
[MentionLeads](https://mentionleads.com/) is an AI customer-discovery and social-listening platform for founders, growth teams, and agencies. Its hosted [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server works with Claude Code, Cursor, and other streamable-HTTP MCP clients.
> Hosted remote server — there is nothing to install or run. This repository contains the public integration docs, configuration examples, and issue tracker.
**Website:** [mentionleads.com](https://mentionleads.com/) · **MCP docs:** [mentionleads.com/mcp](https://mentionleads.com/mcp) · **Endpoint:** `https://mentionleads.com/api/mcp`
## What your agent can do
- Find your newest scored leads and filter them by project or buyer intent.
- Read human-voice Reddit reply and DM drafts for a qualified lead.
- Mine user feedback, competitor complaints, and market pain from real conversations.
- Turn one feedback signal into a scoped, ready-to-run coding task.
- Generate a market-validated implementation plan for a startup idea (Pro).
- Check your MentionLeads plan and daily scan usage.
All MCP tools are read-only. MentionLeads never auto-posts to Reddit, spends a scan, or changes your account through MCP. You review, edit, and post every reply yourself.
## Quick start
### 1. Create an API key
1. [Create a free MentionLeads account](https://mentionleads.com/login?mode=signup).
2. Open **Developer** in the app.
3. Create a key. It starts with `mlk_` and is shown only once.
Keep the key out of source control. Revoke it from the Developer page at any time.
### 2. Connect Claude Code
```bash
claude mcp add --transport http mentionleads https://mentionleads.com/api/mcp \
--header "Authorization: Bearer mlk_YOUR_KEY"
```
Then ask Claude:
```text
What are my highest-intent leads today?
```
### 3. Connect Cursor
Add this server to your MCP configuration:
```json
{
"mcpServers": {
"mentionleads": {
"url": "https://mentionleads.com/api/mcp",
"headers": {
"Authorization": "Bearer mlk_YOUR_KEY"
}
}
}
}
```
A copyable example is available at [`examples/cursor-mcp.json`](examples/cursor-mcp.json). Do not commit a real key.
## Tools
| Tool | Inputs | What it returns |
| --- | --- | --- |
| `list_projects` | — | Your projects, product descriptions, and monitored platforms. |
| `get_leads` | `projectId?`, `minIntent?`, `limit?` | Recent qualified leads with thread URL, intent score, quality, status, and match reasoning. |
| `get_reply_draft` | `leadId` | A suggested public reply and optional DM for human review. |
| `get_product_feedback` | `projectId?`, `limit?` | User feedback, competitor gaps, and market pain found in scanned conversations. |
| `get_build_prompt` | `leadId` | A focused coding task grounded in one real feedback thread. |
| `generate_implementation_plan` | `idea` | A market-validated Markdown implementation plan (Pro, up to 5/day). |
| `get_account_status` | — | Your plan, trial status, and daily scan usage. |
See [the full tool reference](docs/tools.md) for limits, expected inputs, and workflow examples.
## Useful prompts
```text
List my five newest leads with an intent score above 70. Explain why each is worth answering.
```
```text
Find people complaining about my product or a competitor. Pick the strongest signal and turn it into a build prompt.
```
```text
Get the reply draft for this lead. Make it concise and helpful, but do not post it.
```
```text
Validate this idea with real market evidence and create an implementation plan: [describe the problem and audience].
```
## Feedback-to-feature workflow
1. Call `get_product_feedback` to find an actionable complaint or market gap.
2. Pass its `leadId` to `get_build_prompt`.
3. Let your coding agent inspect the codebase and implement the smallest useful change.
4. Call `get_reply_draft`, review the wording, and personally reply to the original thread.
This closes the loop while the conversation is still warm: real feedback becomes a product improvement, and the person who asked for it hears what changed.
## Protocol and security
- Transport: stateless streamable HTTP with JSON-RPC over `POST`.
- Supported protocol versions: `2024-11-05`, `2025-03-26`, and `2025-06-18`.
- Authentication: `Authorization: Bearer mlk_...`.
- Keys are stored by MentionLeads as SHA-256 hashes and are displayed only once.
- Every tool declares MCP's `readOnlyHint` annotation.
- Reply drafts are suggestions for a human to review and post. There is no auto-posting tool.
If a key is exposed, revoke it immediately and create a new one. See [SECURITY.md](SECURITY.md) for private vulnerability reporting.
## Troubleshooting
- **401 / invalid API key:** confirm the header starts with `Bearer mlk_` and create a new key if needed.
- **No leads returned:** create a project, run a scan in the MentionLeads app, or lower `minIntent`.
- **No product feedback returned:** add your brand and competitors to a project watchlist, then run a scan.
- **Method Not Allowed:** the MCP endpoint accepts `POST`; opening it in a browser sends `GET` and correctly returns 405.
More help is available in [the troubleshooting guide](docs/troubleshooting.md).
## About this repository
The MentionLeads MCP server is a hosted product feature. This public repository documents the integration and provides safe client configuration examples; it does not contain the private production server or customer data.
Questions and integration bugs are welcome in [GitHub Issues](https://github.com/borhen68/MentionLeads-MCP/issues). Product support and account questions can be sent through [MentionLeads](https://mentionleads.com/).
## License
The documentation and configuration examples in this repository are available under the [MIT License](LICENSE). Use of the hosted MentionLeads service remains subject to its [Terms](https://mentionleads.com/terms) and [Privacy Policy](https://mentionleads.com/privacy).
