Purpose
reddit-research-mcp is an MCP server that turns Reddit discussions into structured, cited research insights. It is aimed at scenarios such as competitive analysis, customer discovery, and market research. It is an open-source component of the Dialog research engine and can also be used completely independently and for free.
Main capabilities
- Evidence-based insights with full citations: every finding links back to a real Reddit post or comment, with vote counts, awards, and direct URLs.
- Zero-friction setup: no Reddit API credentials, no terminal commands, no credential management; just connect and start researching.
- Large-scale semantic search: the Reddit API search limit is 250 results; this server uses vector embeddings to perform conceptual search across 20,000+ indexed subreddits, discovering relevant communities you did not know about.
- Persistent research management: save subreddit collections as feeds for continuous monitoring, suitable for long-term competitive analysis and market research activities.
Usage
Quick setup (about 60 seconds), supporting OpenAI Codex CLI, etc. The MCP server URL is https://mcp.dialog.tools/mcp，适用于其他 AI assistant. Can be used in Claude Code, Cursor, Codex, Gemini CLI, or any MCP-compatible client. Uses Descope OAuth2 for secure authentication: no Reddit credentials required, tokens are managed automatically by the MCP client, and only public Reddit data is accessed; first-time authentication takes about 30 seconds.
Server operations include: discover_subreddits (semantic vector search across 20,000+ indexed subreddits to find relevant communities), search_subreddit (search posts within a specific subreddit by time range and sort order), list posts from a single subreddit by type (hot, new, top, rising), batch-fetch posts from multiple subreddits concurrently (70% efficiency improvement), and retrieve full comment trees for in-depth analysis.
Feeds are used to save research configurations for continuous monitoring: create_feed (save discovered subreddits along with analyses and metadata), list_feeds (view all saved feeds with pagination), get_feed (retrieve a specific feed by ID), update_feed (modify a feed's name, subreddits, or analysis), delete_feed (permanently delete a feed).
Use cases
By role:
- Indie developers and SaaS founders: validate product ideas before building; find the communities where target customers gather; monitor competitor mentions and sentiment; discover unmet needs in niche areas.
- Product managers: gather customer feedback at scale; track feature requests across communities; understand the competitive landscape; identify emerging trends before they peak.
- Market researchers: conduct sentiment analysis with full citations; build audience personas from real discussions; track how opinions evolve over time; generate evidence-based reports.
Specific uses include competitive analysis (compare sentiment, feature requests, pain points, and migration experiences, with links to every discussion), customer discovery (discover unmet needs, feature gaps, and pricing concerns from the target market), market research (track adoption trends, concerns, success stories, and emerging use cases, with temporal analysis), product validation (identify pain points from real Reddit discussions and validate solutions), and continuous monitoring (build curated feeds of important communities).
Limitations
- Only accesses public Reddit data.
- The Reddit API itself has a search limit of 250 results, but this server expands coverage through semantic search.
- If you needIt runs automatically as scheduled and pushes to Slack, which requires integration with the Dialog platform; the server itself does not have this scheduling capability when used independently.
- Specific pricing information is not yet available.
Technical Details
Tech stack: Python 3.11+ (with type hints), FastMCP as the server framework, ChromaDB for semantic search, and PRAW for Reddit API interaction. The server follows a layered abstraction pattern to improve extensibility and self-documentation: Layer 1 discovery (view available operations and get workflow suggestions), Layer 2 schema inspection (understand parameter requirements, validation rules, and examples before execution), Layer 3 execution (perform actual operations using validated parameters).
Version 1.0.1. Contributions are welcome; issues can be submitted in the repository.
