trendzeist-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| TRENDZEIST_HL | No | UI language for Google Trends. | en-US |
| TRENDZEIST_TZ | No | Timezone offset in minutes. | 360 |
| TRENDZEIST_BACKOFF | No | Exponential backoff factor. | 1.5 |
| TRENDZEIST_PROXIES | No | Comma-separated proxy URLs for rotation. | |
| TRENDZEIST_RETRIES | No | Retry attempts on transient errors. | 3 |
| TRENDZEIST_CACHE_DIR | No | Persistent cache location; use 'off' to disable. | OS user cache dir |
| TRENDZEIST_LOG_LEVEL | No | Python logging level (stderr). | WARNING |
| TRENDZEIST_MIN_INTERVAL | No | Minimum seconds between Google requests. | 2.0 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| interest_over_timeB | Search-interest time series (0-100) for 1-5 keywords, with per-keyword summary (mean, latest, peak, direction, growth_3m/growth_12m % and a plain-English insight). Long series are downsampled to ~60 points. gprop: '' (web), 'news', 'youtube' (authority channels answer engines cite), 'images', 'froogle'. |
| compare_keywordsA | Compare 2-5 keywords head-to-head: relative share, leader and trend direction for each. Use to pick the strongest angle among alternatives. |
| related_queriesA | Top and rising search queries related to a keyword. 'rising' includes %
growth and is_breakout (>5000%); every item carries a title |
| related_topicsA | Top and rising Knowledge-Graph topics related to a keyword. Google often returns nothing here; if available=false fall back to related_queries. |
| interest_by_regionB | Where a keyword is searched most. resolution: COUNTRY (geo='' only), REGION (states/provinces within geo), CITY and DMA (US metro areas; geo='US' or worldwide only). Non-US countries support REGION only. |
| suggest_keywordsA | Google's entity suggestions for a keyword (title, type, mid). Use to disambiguate ambiguous terms or find the canonical topic id. |
| trending_nowA | Real-time trending searches (last ~24h) for a country or US state, with approximate traffic and linked news headlines. Good for newsjacking. |
| list_categoriesA | Browse/search Google Trends category ids (e.g. 'coffee' -> Food & Drink >
Coffee & Tea). Pass the id as |
| discover_topicsA | One-shot topic discovery for blog ideation. For 1-5 seed keywords, pulls
trend direction plus rising/top related queries, de-duplicates and ranks
candidates as breakout > rising > evergreen, each tagged with a title |
| mine_questionsA | Long-tail questions people type about a seed, from Google Autocomplete ('how to', 'why', 'what is', 'vs' ... expansions), de-duplicated and tagged with a title angle. Feeds FAQ sections and answer-engine (AEO) content. Up to ~16 throttled requests per call; results cached 24 h. limit: 1-100. |
| aeo_opportunitiesB | AEO opportunity finder. For 1-5 seeds: question-shaped searches (Trends + a
capped Autocomplete pass) clustered by seed and title angle, each scored with
seed trend direction, Google News coverage (who to quote / pitch), Wikipedia
presence and the evidence type that makes the answer citable. Up to ~1+10
requests per seed; partial failures per source in |
| news_coverageA | Who is covering a topic in Google News: recent headlines, a publisher frequency table (mention / pitch targets) and a recency histogram with a coverage label. One request, cached 15 min. limit = headlines returned (1-50). |
| wiki_attentionA | Does Wikipedia have an exact-title article for a topic, and is attention growing?
Exact article (title, encoded URL, wordcount), daily pageviews for |
| trendzeist_statusA | Health of every data source (google_trends, google_autocomplete, google_news, wikipedia: live / error / idle with the last error), cache usage and effective settings. Free: makes no network requests. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| blog_ideas_from_trends | Guided workflow: turn Google Trends data into a prioritised list of blog post ideas. |
| answer_brief | Citable-answer brief for one question: direct answer, statistic, quote, sources, FAQ, schema (GEO structure). |
| content_brief | Full content brief for a topic: titles, H2 outline, FAQ, regions, evidence checklist, publishers to pitch. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 14 tools
Most tools target distinct data sources (Trends time series, News, Wikipedia, Autocomplete, Geo). However, the discovery-oriented tools (aeo_opportunities, discover_topics, related_queries, mine_questions) all produce content angles/questions and genuinely overlap, forcing the agent to rely on descriptions to choose. Boundaries are mostly clear but not perfectly crisp.
All names use consistent snake_case, which is good. There's a mild mix between verb_noun (mine_questions, suggest_keywords, list_categories, compare_keywords) and noun_phrase (interest_over_time, related_queries, trending_now, news_coverage), but the pattern is predictable and readable throughout.
14 tools sits squarely in the well-scoped 3-15 range. Each tool maps to a distinct data source or analysis task (trends, news, wiki, autocomplete, geo, categories, status), so no tool feels redundant or padded.
The SEO/AEO research domain is well covered: trend direction, related queries/topics, questions, regional interest, categories, news, Wikipedia, and health status. Minor gaps exist (no direct cross-keyword or cross-region time comparison beyond compare_keywords/interest_by_region), but core workflows have no dead ends.