GitHub daily star history: an MCP database you can query from Claude Code or Cursor
About 850,000 repositories with 50+ stars, with star counts recorded daily: 1-day, 7-day and 30-day gains, and whether a repo just emerged.
What you can ask
- Which open-source projects gained the most stars this week, and which are brand new?
- When did this repo start growing, and are its stars suspicious?
- What new projects are rising in music generation?
These are example phrasings. Once connected, ask in plain language; the agent decides which table to query and how to filter.
Example: a real query and what it returned today
Your agent calls the MCP data tool like this (you never write it yourself — you just ask):
data(lib="github", args="rising --days 7")
Repositories with the largest 7-day star gain, first 8 rows:
| full_name | stars | gain_7d | status | age_days | language | suspect |
|---|---|---|---|---|---|---|
| morluto/rea | 449543 | 23090 | 稳定增长 | 211 | TypeScript | 单日涨星超 2 万 |
| storytold/craft-anything | 25993 | 18237 | 新冒出来 | 11 | Rust | |
| sherlock-project/sherlock | 108668 | 13181 | 加速 | 2879 | Python | |
| pingdotgg/t3code | 55496 | 11526 | 稳定增长 | 276 | TypeScript | |
| ccch1mneyyy/dsh-Go | 18619 | 11285 | 加速 | 17 | Dart | |
| cloudflare/security-audit-skill | 69016 | 10943 | 稳定增长 | 146 | JavaScript | |
| ClickHouse/ClickStack | 10016 | 9835 | 加速 | 364 | Shell | |
| Ebony-Vinyl/dsh-our-free-model | 68649 | 9745 | 稳定增长 | 48 | JavaScript |
How fresh the data is
This is today’s output of coverage (row counts and date ranges per table):
| t | source | n_rows | first_date | last_date | n_repos |
|---|---|---|---|---|---|
| built_at | 2026-11-11T03:47 | ||||
| repo | census | 865543 | 865543 | ||
| repo | deeply | 1072 | 1072 | ||
| repo_profile 有 gain_1d | 860053 | ||||
| repo_profile 有 gain_30d | 16510 | ||||
| repo_profile 有 gain_7d | 856190 |
Limits: what this data cannot tell you
Continuous daily snapshots start on 2026-10-29; earlier points are backfilled. Stars can be bought — suspicious repos are flagged, but not exhaustively.
It is a remote MCP server: sign in with Google, paste one prompt and your agent configures itself, then ask in plain language inside your project. Free during beta: 200 queries a day, up to 50 rows each.
Install the skillOr just ask on the web →FAQ
How do I query it from Claude Code, Cursor or Codex?
It is one library inside the Stringers remote MCP server. Sign in with Google at stringers.io/en/start.html to get a personal URL, paste the prompt to your agent and it configures itself; then just ask — the agent calls the data tool with lib "github".
Is it free?
Free during beta: 200 queries per account per day, up to 50 rows per query.
Workflows that use this database
Other databases
- Verified indie product revenue
- Small online businesses for sale
- App Store charts in 115 regions
- Google Play installs and charts
- Mac App Store charts
- Chrome Web Store extensions
- Shopify App Store
- Product Hunt launches
- Hacker News stories
- Meta Ad Library tracker
- Google Ads Transparency tracker
- TikTok ad library tracker
- Daily breakout list
- Posts from five X circles