GitHub New Repo, Star-Jump & Release Alerts (API)

Find new and fast-rising GitHub repositories by topic, language and stars, and get new releases of the repositories you watch, as clean JSON rows. It uses the official GitHub REST API (no page scraping, no login). Schedule it daily and it returns only what's new since the last run: new repos, new releases, and repos whose star count jumped.

Built and maintained by FedSignals Data.

What it does

• Discovers new repos: searches GitHub for repositories created in the last N days with at least N stars, for each topic you choose (e.g. llm, mcp, ai-agents), optionally limited to languages (e.g. Python, Rust).
• Tracks releases: for every owner/name in watchRepos, returns new releases with tag, date, link and release notes.
• Flags star jumps: on scheduled runs, a repo you already received comes back as star_jump when it gained at least starJumpThreshold stars since the previous run.
• Incremental by default: onlyNew remembers what it already delivered (per filter set), so a daily schedule returns, and charges for, new rows only.

Use cases

• Daily "new AI / MCP / agent repos" feed for a newsletter, Slack channel or market-research sheet
• Release alerts for the libraries and tools your product depends on
• Spotting breakout open-source projects early (VCs, devrel, competitive intelligence)
• Feeding an LLM agent or RAG pipeline with fresh repos and changelogs

Input

Field	Type	Default	What it does
topics	list	["mcp"]	GitHub topics to search, one search per topic. Empty = all topics.
languages	list	any	GitHub language names, e.g. Python, TypeScript, Go. Misspellings fail before any row is charged, with a suggestion.
minStars	integer	50	Minimum stars.
createdWithinDays	integer	30	Only repos created in the last N days (1–365).
watchRepos	list	–	owner/name or github.com URLs to watch for new releases (up to 200).
onlyNew	boolean	true	Return only rows not delivered by an earlier run with the same filters, plus star jumps.
starJumpThreshold	integer	100	Stars gained since the last run that trigger a star_jump row (0 = off).
maxItems	integer	100	Max rows (and charges) per run.
githubToken	secret string	–	Optional, your own token. Raises GitHub limits (see Limits). Stored encrypted, never logged.
discoverRepos	boolean	true	Turn off to track watchRepos releases only.
fetchLatestRelease	boolean	true	Adds each discovered repo's latest release (one extra request per repo).
releasesPerRepo	integer	1	Releases per watched repo on the first run or with onlyNew off (1–10).
includePrereleases	boolean	false	Also return alpha/beta/rc pre-releases.
includeForks	boolean	false	Also return forks in discovery.
stateKey	string	hash of filters	Separate memory per saved task.

Example input:

{
  "topics": ["mcp", "ai-agents"],
  "languages": ["Python", "TypeScript"],
  "minStars": 100,
  "createdWithinDays": 30,
  "watchRepos": ["modelcontextprotocol/python-sdk", "ollama/ollama"],
  "onlyNew": true,
  "maxItems": 10
}

maxItems 10 caps this run at 10 rows × $0.002 = $0.02, plus the $0.00005 start fee ($0.02005 in total). If you cap cost with Apify's run spending limit instead, the start fee counts toward it: a $0.02 limit stops at 9 rows, so set $0.021 to get all 10.

Watch releases of projects from outside the US, e.g. Mistral AI (France) and DeepSeek (China) (tested Oct 30, 2026):

{
  "discoverRepos": false,
  "watchRepos": ["mistralai/mistral-inference", "deepseek-ai/DeepSeek-V3"],
  "onlyNew": true,
  "maxItems": 50
}

Output

Each dataset row is one event:

{
  "eventType": "new_release",
  "repo": "modelcontextprotocol/python-sdk",
  "url": "https://github.com/modelcontextprotocol/python-sdk",
  "description": "The official Python SDK for Model Context Protocol servers and clients",
  "topics": ["mcp", "mcp-client", "mcp-server", "python"],
  "language": "Python",
  "stars": 24711,
  "starsGainedSinceLastRun": null,
  "forks": 4092,
  "openIssues": 378,
  "createdAt": "2024-09-24T21:01:35Z",
  "pushedAt": "2026-10-27T11:15:12Z",
  "homepage": "https://py.sdk.modelcontextprotocol.io/",
  "license": "MIT",
  "isFork": false,
  "latestRelease": {
    "tag": "v2.3.0",
    "name": "v2.3.0",
    "publishedAt": "2026-10-02T22:02:27Z",
    "url": "https://github.com/modelcontextprotocol/python-sdk/releases/tag/v2.3.0",
    "prerelease": false
  },
  "matchedTopic": null,
  "releaseNotes": "`pip install -U mcp`. Docs: ... (max 1,500 characters)",
  "scrapedAt": "2026-10-30T16:09:20Z"
}

• eventType: new_repo, new_release or star_jump
• starsGainedSinceLastRun: filled on incremental runs once a repo has been seen before
• latestRelease: for new_release rows this is the new release; null when the repo has no releases
• releaseNotes: the release notes on new_release rows, with @mentions and e-mail addresses removed; null on new_repo and star_jump rows
• All dates (createdAt, pushedAt, latestRelease.publishedAt, scrapedAt) are ISO 8601 in UTC (Z), whatever the repo owner's time zone. There are no money fields.

A SUMMARY record in the run's key-value store lists the searches and their total matches, rows per event type, rows skipped as already seen, watched repos with no published releases (watchedWithoutReleases), GitHub rate-limit state and why the run stopped (completed, maxItems, spendingLimit, rateLimited, upstreamError).

Pricing

Pay per event, no subscription:

Event	Price
Repository row (new_repo or star_jump)	$0.002
Release row (new_release)	$0.002
Actor start	$0.00005

That is $2 per 1,000 rows. A daily run that finds 20 new repos and 2 releases costs about $0.044. Already-seen rows are skipped and never charged. Invalid input fails before any row is charged (only Apify's $0.00005 start fee applies). Set a maximum cost per run in Apify and the Actor stops cleanly when it is reached.

FAQ

Do I need a GitHub token? No. Without one, GitHub allows 10 searches a minute and 60 other requests an hour per IP, which covers small runs (a few topics, ~25 latest-release lookups, a handful of watched repos). For many topics, many watched repos or frequent runs, add your own token in githubToken; a fine-grained token with no extra permissions is enough. It is only sent to api.github.com.

Why did I get fewer rows than GitHub shows? With onlyNew on, rows already delivered by earlier runs with the same filters are skipped (see skippedAlreadySeen in SUMMARY). Turn onlyNew off to get every match.

How do I get alerts in Slack or e-mail? Save a task, schedule it daily, and add an Apify integration (Slack, e-mail, webhook, Zapier, Make) on run success.

Does it collect personal data? No profile data is collected: no names, e-mails, followers or contributor lists. Repository names contain the owner's public login, as on GitHub. Release notes have @mentions and e-mail addresses removed.

What happens when GitHub's rate limit is hit? The Actor waits up to about a minute; if the limit is longer, it returns what it has collected, says so in the status message and SUMMARY, and you are only charged for delivered rows.

Other languages and regions

Works for public repositories from any country. topics and languages use GitHub's own names (topics are usually in English); repo descriptions are returned as written.

Also searched as: repositorios en tendencia de GitHub, seguimiento de versiones (ES) · repositórios em alta no GitHub, acompanhamento de versões (PT) · dépôts GitHub tendance, suivi des versions (FR) · GitHub-Trends, Release-Tracker (DE).

Limits

• GitHub's search API returns at most 1,000 results per query; narrow the filters (more stars, fewer days, a language) for broad topics.
• Topics in one run are searched separately (GitHub combines multiple topics with AND); languages are combined with OR.
• Without a token, up to 25 latest-release lookups per run; the rest are counted in releaseLookupsSkipped.
• Star jumps compare against the previous run with the same filters, so they need a schedule.
• Watched repos with no published GitHub releases (only tags, or nothing yet) return no release rows; the status message names them and nothing is charged for them.
• Private repositories are not supported.

Source and terms

Data comes from the public GitHub REST API (search, repositories and releases endpoints), used within GitHub's rate limits. Use of the data is subject to the GitHub Terms of Service (section H, API Terms) and Acceptable Use Policies; don't use it for spam or to sell personal information. This Actor is not affiliated with or endorsed by GitHub.

Support

Questions or feature requests: open an issue on the Actor's Issues tab or e-mail fedsignals.alerts@gmail.com.