Dialog MCP Server
An MCP server for Reddit research: semantic community search and post and comment reading with links to sources
- 265 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Reddit queries pass through the developer's server, which can see your research topics
- Feeds with community sets and analysis notes are stored on the service side
- During sign-in, the server processes the token and identity data supplied by the Descope OAuth provider
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/dialog-mcp-serverDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Reddit queries pass through the developer's server, which can see your research topics
- Feeds with community sets and analysis notes are stored on the service side
- During sign-in, the server processes the token and identity data supplied by the Descope OAuth provider
This is third-party code. Review the repository files before installing.
What it does
Dialog MCP Server connects an agent to a hosted Reddit research service. It finds communities from a topic description, selects relevant posts and reads discussions. Results include links so you can check conclusions against the original comments. The server exposes capabilities progressively. The agent first discovers operations, retrieves the parameter schema for one operation and then executes it. This combines community search, posts and comments without keeping a long list of tools in context. You can save community selections as feeds with analysis notes and revisit them in another session. Feeds support creation, reading, editing and deletion. A built-in prompt provides a repeatable research workflow; the MCP server does not deliver scheduled reports.
Who it is for. For founders, product managers and market researchers who want to validate ideas and competitors from Reddit discussions right inside Claude Code, Cursor or Codex.
Good fit when
- You want to compare what developers say about two tools and get a summary of complaints and requests with links to the threads
- You are testing a product idea and looking for pain points raised in niche communities, for example around subscription billing
- You need to find target audience communities from a topic description when you do not know the exact name
- You want to save a set of communities as a feed and return to it in later sessions
- The answer needs quotes with vote counts and direct URLs to paste into a report
Not a fit when
- You need to publish posts or write comments on Reddit: the server operations only read and manage feeds
- You need direct messages, private communities or your account data: the server sees public material only
- Research should run on a schedule with a report in Slack or Telegram: the Dialog platform does that, while the free server works on request inside a session
- You need full coverage of small subreddits: the index starts at communities of 2,000 members
Example request
Find communities that discuss CRMs for small business, collect the main complaints with quotes and links, and save the communities as a feedMCP
- Transport
- http
- Authentication
- OAuth
Limitations
Queries and saved feeds pass through Dialog infrastructure, with Descope OAuth2 handling sign-in. Responses contain public Reddit content; research topics and authentication data are still user data. Avoid confidential information in queries.
The author describes an index of communities with at least 2,000 members, refreshed weekly. It does not guarantee complete Reddit coverage. On October 31, 2026 the public MCP is described as free, but check limits and availability with its operator. Self-hosting requires Reddit API credentials, Descope, an index and feed storage; Python and uv alone are insufficient.
How to disable. Run `claude mcp remove dialog-mcp` in Claude Code, or delete the dialog-mcp block from the Cursor, Codex or Gemini CLI config. Saved feeds live on the service side; delete them with the `delete_feed` operation before you disconnect.
FAQ
How do I connect Dialog's Reddit MCP to Claude Code?
Run `claude mcp add --scope local --transport http dialog-mcp https://mcp.dialog.tools/mcp` and sign in through OAuth on the first call. For Codex and Gemini CLI the repository gives `mcp-remote` commands, and Cursor has a one-click install link.
Do I need Reddit API credentials?
The hosted server needs no Reddit API credentials; sign in through Descope OAuth2. A self-hosted instance requires Reddit API credentials and the other services configured. Both options use public Reddit data.
Why does the server have only three tools?
It is built in three layers: `discover_operations` lists operations, `get_operation_schema` describes parameters and examples, `execute_operation` runs them. The agent does not have to keep ten separate tool descriptions in context.
What are feeds and what are they for?
A feed stores a set of discovered subreddits with analysis and metadata. The `create_feed`, `list_feeds`, `get_feed`, `update_feed` and `delete_feed` operations let you come back to the set in a new session and track competitors over time.
Can I host Dialog MCP myself?
The source uses MIT, but self-hosting requires Reddit API and Descope credentials plus search and feed storage. The hosted connection avoids that setup for the user; it is a separate deployment option.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server with SEO data: backlinks, keyword difficulty and traffic estimates from the free Ahrefs tools
Humblytics MCP gives an agent live site analytics, funnels and A/B tests plus 12 marketing skills for finding conversion problems.
DataForSEO MCP filters API responses to selected fields, so an agent receives only the data needed for the current analysis.
Remote Analook MCP server builds a competitor report from a site URL: traffic, social footprint, Product Hunt and AI insights.
DataLikers MCP
DataLikers MCP Server
DataLikers MCP server for Instagram data: profiles, posts, hashtags, locations, stories and music
Remote ThoughtSpot Spotter MCP server: an agent asks questions about company data and gets live answers with links
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Dialog MCP Server (this card) | no | OAuth | 265 | |
| Reddit MCP Buddy | no | not required | 894 | reads Reddit posts and analyzes user activity, with no semantic community index |
| MCP Reddit Server | no | not required | 438 | returns the hot threads of a subreddit together with comments |
| reddit-research-skills | no | 15 | a set of Reddit research skills instead of a remote MCP server |
README in short
The server uses Python and FastMCP, with PRAW accessing Reddit in read-only mode. Example research reports in the repository show how findings can be presented.