Gentic Research — Documentation

Give your AI agent live access to consumer conversations and search trends. Mine Reddit for sentiment, pull Google Trends interest curves, and store every run for later — all through the Model Context Protocol.

1. Getting Started

Sign Up & Get Your API Key

Before you can use Gentic Research, you need an API key to authenticate your requests.

  1. Go to gentic.co/research and create an account.
  2. Create an organization from your dashboard. API keys and billing are scoped to the organization.
  3. Generate an API key and use it as a Bearer token in your MCP client.

2. Connecting to the MCP Server

The server is available at https://mcp.gentic.co/research. For Claude Code:

claude mcp add gentic-research \
  --transport http \
  https://mcp.gentic.co/research \
  --header "Authorization: Bearer YOUR_API_KEY"

For Claude Web and ChatGPT you can also connect via OAuth — no API key needed. See the connect section on the landing page for other MCP clients (n8n, OpenClaw).

3. Agent Skill

For the best results, pair the MCP server with the Gentic Research agent skill. The MCP server gives your agent tool access; the skill teaches it the optimal workflow order. Both the raw SKILL.md and a ready-to-upload .skill bundle are generated on demand from the live manifest, so they always reflect the current tools and pricing.

Add the skill directly via URL:

https://gentic.co/research/SKILL.md

Or upload a .skill bundle to Claude Managed Agents:

https://gentic.co/research/gentic-research.skill

Download this file and upload it wherever Claude Managed Agents asks for a .skill file. It's a zip bundle generated on demand from the latest SKILL.md.

4. When to Apply

  • User wants to investigate consumer sentiment, problems, or discussions around a topic, product, or category.
  • User asks to compare keyword interest over time, across regions, or spot emerging trends.
  • User wants to replay or retrieve a previous research run by run_id, or list recent runs.
  • User is building a research brief and needs both qualitative (Reddit) and quantitative (Google Trends) signals in one pass.

5. Workflow

  1. 1. Pick the right tool for the question

    Qualitative 'what are people saying / complaining about' questions → `search_reddit`. Quantitative 'how popular is X / is X growing' questions → `google_trends`. Retrieval or replay → `get_research_results`. When both qual and quant are useful, run the two in parallel.

  2. 2. Write descriptive Reddit queries and tune `limit` carefully

    `search_reddit` rewards specificity. 'hair loss treatment side effects' beats 'hair loss'. If the user names a community, pass up to 5 subreddits via `subreddits`; otherwise search globally. Use `sort=top` + `timeFilter=month` for signal, `sort=new` for recency. **Every Reddit result costs 5¢** — default `limit=25` is a good starting point (≈$1.25), bump to 50 for broad scans (≈$2.50), drop to 10 for targeted intelligence. Mention the cost to the user when running >25 results.

  3. 3. Compare keywords on Google Trends

    `google_trends` works with a single keyword, but it shines with 2–5 keywords compared side by side — that's when you get relative interest curves instead of a flat absolute number. Default `timeRange=12m`, widen to `5y` for long-term shifts. Pass `includeRegions=true` when the user cares about geography.

  4. 4. Save the run_id

    Every `search_reddit` and `google_trends` call returns a `run_id`. Capture it. Later the agent (or a human) can call `get_research_results` with that id to retrieve the exact same data for free — useful for citations, briefs, or feeding another MCP server.

  5. 5. Present results clearly

    Don't dump raw JSON. For Reddit: group posts by theme, render a ranked table (title · subreddit · score · link), and quote verbatim user language — that's the whole point of Reddit research. For Google Trends: lead with the overall pattern (flat / rising / seasonal / comparative winner), then call out **rising related queries** as leading indicators. Always end with one concrete next step the user can take (another search, a chained MCP call, a brief).

  6. 6. Chain with other Gentic servers

    The agentic win is composition: turn Reddit pain points into creative briefs via Gentic Creative, turn Google Trends winners into campaign keywords for Gentic Meta, or pipe either into Gentic Data for structured storage. The MCP surface is the same across all of them.

6. Tool Reference

3 tools, rendered live from the Gentic MCP manifest. Parameter tables come directly from each tool's JSON Schema.

get_research_results

Free

Retrieve stored Reddit research results. Pass a run_id to get full results from a specific search, or omit it to list recent research runs with metadata.

ParameterTypeDescription
run_idstring

Specific research run ID to retrieve. Omit to list recent runs.

limitnumber

Max results to return. Defaults to 25.

1 – 100

search_reddit

5¢ / result (min 5¢)

Search Reddit for consumer discussions, problems, and sentiment around a product category or topic. Returns structured posts with title, body, score, subreddit, and URL. Results are stored for later retrieval. Billed per result.

ParameterTypeDescription
query
required
string

Search query — be descriptive for better results, e.g. 'hair loss treatment side effects' rather than 'hair loss'

subredditsstring[]

Filter to specific subreddits (max 5), e.g. ['tressless', 'HaircareScience']. Omit for global search.

sortstring

Sort order. Defaults to 'relevance'.

enum: relevance, hot, new, top, comments

timeFilterstring

Time filter for results. Defaults to 'all'.

enum: hour, day, week, month, year, all

limitnumber

Number of results to return (1-100). Defaults to 25.

1 – 100

7. Pricing

Pricing is pulled live from the Gentic MCP manifest. All prices are per call and deducted from your Gentic credits.

ToolCost
get_research_resultsFree
google_trends10¢ / call
search_reddit5¢ / result (min 5¢)

8. Notes

  • All runs are organization-scoped and persisted in MotherDuck — users only see their own runs.
  • `search_reddit` is the only billed tool at 5¢ per result. At 25 results that's $1.25; at 100 it's $5. Mention the cost when running large pulls. `google_trends` and `get_research_results` are free.
  • Reddit is routed through ScrapingBee so it works from cloud IPs. Subreddits that return 403/404 are silently skipped instead of failing the whole call.
  • Rate limit: 100 requests per minute per API key. Contact support@gentic.co for higher limits.