MCP Server Trend Radar

MCP Server Trend Radar is a daily / weekly radar of the fastest-growing Model Context Protocol (MCP) servers on GitHub. Instead of ranking by absolute star count like a static catalog, it measures momentum: how many stars each MCP server gained in the last 7 / 30 / 90 days. Find the next breakout MCP integration before it's saturated.

This Actor was built by Ian Mu as Actor #12 in a 100-actor portfolio. It pairs naturally with the companion mcp-server-catalog Actor (point-in-time quality scoring) and with the verification harness at github.com/ianymu/claude-verify-before-stop.

What does MCP Server Trend Radar do?

It scans GitHub for repos tagged with the mcp and model-context-protocol topics that have been actively pushed inside your chosen window, then samples each repo's stargazers timeline (via the paginated application/vnd.github.v3.star+json accept header) to estimate stars gained in the last N days. It computes starsPerDay and a normalized starsVelocityScore (0-100, anchored to the top mover in the run), then ranks the results.

Run it on the Apify platform to get free scheduling (daily 09:00 UTC works well), webhook delivery into Slack / Notion / your newsletter pipeline, dataset versioning, and a managed proxy pool if you ever need it. You can hit the dataset via the Apify API and pipe it straight into a curated newsletter or an investor dashboard.

Why use MCP Server Trend Radar?

• Founders & devs spot new MCP integrations early, before everyone wraps them.
• Investors track which MCP ecosystem niches (databases, agents, devops, design) are heating up week-over-week.
• Newsletter writers / curators auto-generate a "this week in MCP" section without hand-curating GitHub.
• MCP server authors see who's beating them on momentum and study what's working.

Unlike mcp-server-catalog, which scores quality at a single point in time (stars + recency + license + activity), this Actor is purely a momentum signal. A repo with 50 stars and +40 stars last week will outrank a 5,000-star repo that's been flat. Run both Actors weekly and compare quality vs. velocity to find rising stars worth tracking.

How to use MCP Server Trend Radar

1. Click Try for free in the Apify Console.
2. Set windowDays (default 30) — drop to 7 for a weekly radar, or push to 90 for a quarterly view.
3. Set maxResults (default 30).
4. Optionally extend topicFilter with ["mcp", "model-context-protocol", "llm-tools"] if you want broader coverage.
5. Click Start and let the Actor run. A first run on default settings completes in roughly 60-180 seconds depending on GitHub rate limits.
6. Open the Dataset tab to see ranked results, or download the dataset as JSON / CSV / Excel. A Markdown leaderboard is also written to the key-value store as mcp-trend-radar.md.

Input

Field	Type	Default	Description
windowDays	integer	30	Window over which to measure star velocity (1-365).
maxResults	integer	30	Maximum repos to return (1-200).
topicFilter	string[]	["mcp", "model-context-protocol"]	GitHub topics that define "an MCP server" for the radar.

Example input JSON:

{
    "windowDays": 7,
    "maxResults": 25,
    "topicFilter": ["mcp", "model-context-protocol"]
}

Output

Each row in the dataset looks like:

{
    "repo": "supabase/mcp-server",
    "url": "https://github.com/supabase/mcp-server",
    "currentStars": 1250,
    "starsGainedInWindow": 380,
    "starsPerDay": 12.7,
    "starsVelocityScore": 95,
    "description": "Supabase Model Context Protocol server",
    "language": "TypeScript",
    "lastPushed": "2026-05-18T10:23:00Z",
    "windowDays": 30
}

You can download the dataset in various formats such as JSON, HTML, CSV, or Excel. Rows are sorted by starsVelocityScore descending.

Data table

Field	Meaning
repo	owner/name of the GitHub repo.
url	Direct link to the repo.
currentStars	Star count at the moment the Actor ran.
starsGainedInWindow	Estimated stars gained in the last windowDays days.
starsPerDay	starsGainedInWindow / windowDays, rounded to one decimal.
starsVelocityScore	0-100, normalized against the fastest grower in this run.
description	GitHub repo description.
language	Primary language reported by GitHub.
lastPushed	Last push timestamp.
windowDays	The window setting used for this row (useful when combining runs).

Pricing / Cost estimation

The Actor is lightweight: a typical 30-result run on default settings uses well under one compute unit on the Apify free tier. Most of the work is GitHub API calls; the stargazer-sampling binary search caps probes at six pages per repo, so the worst case for maxResults=30 is roughly 60 search calls + 360 stargazer calls, comfortably inside the unauthenticated GitHub rate limit when run hourly or less, and totally fine with a token at any cadence.

If you set GITHUB_TOKEN (or GH_TOKEN) in the Actor's environment variables, the radar will use it automatically and lift the GitHub rate limit from 60 to 5,000 requests per hour.

Tips and advanced options

• Weekly newsletter: windowDays: 7, maxResults: 10, schedule daily at 09:00 UTC, and pipe the output into your newsletter via Apify webhook.
• Investor scan: windowDays: 30, maxResults: 100, schedule weekly, and join with mcp-server-catalog output to compare velocity vs. quality.
• Niche radar: extend topicFilter with niche topics (e.g. ["mcp", "claude-mcp", "openai-mcp"]) to slice the ecosystem.
• Speed: provide a GITHUB_TOKEN env var to skip rate-limit pauses entirely.

FAQ, disclaimers, and support

Is the star velocity exact? It's an estimate. The GitHub stargazers API is paginated and per-page sampled, so very large repos (>50k stars) are approximated via binary search rather than fully scanned. The accuracy is typically within +/-10 stars over a 30-day window for repos under 10k stars — more than enough to rank momentum.

Why might my favorite MCP server be missing? It needs the mcp or model-context-protocol GitHub topic to appear in the candidate pool. Open a PR on the upstream repo to add the topic, or extend topicFilter to include topics the project actually uses.

Legal: this Actor only reads public GitHub data via the official GitHub REST API. No scraping of behind-auth or rate-limit-evasion behavior.

Issues / feature requests: open an issue against this Actor on Apify, or ping @ianymu on GitHub. Custom variants (weekly digest delivery, Notion sync, multi-ecosystem radars for LangChain / LlamaIndex / agent frameworks) are available on request.

Built by Ian Mu. Quality verification harness: github.com/ianymu/claude-verify-before-stop.