GitHub Trending Discovery - Daily Repositories & Changes

GitHub Trending Discovery

Find repositories appearing on GitHub's daily Trending lists. Choose a language, set a small result limit and get ranked observations with exact star counts, stars gained today, source links and collection timestamps.

Use it to build developer newsletters, discover emerging AI projects, refresh engineering research watchlists or feed a repository-review workflow. Optional saved comparisons show how a repository's rank and total stars changed between your own successful observations.

Start in one minute

1. Keep Languages set to python, or use a GitHub language slug such as typescript, rust or go. Use all for the all-language list.
2. Set Maximum results to 5 for a small first run.
3. Run the Actor and open the Repositories table. Export JSON, CSV, Excel or a dataset URL from Apify.

No GitHub token, browser cookies or proxy configuration is required. The Actor reads public daily Trending pages and checks their robots policy. A source failure is reported in SUMMARY; failed pages do not produce substitute records.

{
  "languages": ["python"],
  "maxResults": 5
}

What one row means

One row is one repository observed on one language's daily Trending list at observedAt. The same repository on two selected lists produces two observations, because its rank and list context differ.

Field	Meaning
fullName	Exact source owner/repository name
rank	Position in the complete fetched daily list, before your filters
listLanguage	Selected list slug, or all
programmingLanguage	Main language displayed on the repository card; null if absent
stars / forks	Exact integer counts displayed by GitHub; missing values stay null
starsToday	GitHub's displayed “stars today” count; not a calculation from stored history
description	Public repository description, when present
url / sourceUrl	Repository and exact Trending list URLs
observedAt	UTC time this response was parsed
recordId	Stable identity for a repository within a particular language list
comparisonStatus	not_requested, new, changed or unchanged
rankChange	Previous observed rank minus current rank; positive means a higher placement
starsChange	Current total stars minus the previous observation's total stars
changedFields / previous	Changed field names and the prior saved observation, when comparison is enabled

rankChange is not an official GitHub historical metric. Lists change continuously; observations made hours apart are not a reconstruction of yesterday's final ranking. GitHub's “today” window is source-defined. Star counts can fall, so negative changes are preserved.

Useful recipes

A daily engineering discovery list

{
  "languages": ["typescript"],
  "maxResults": 10
}

Find AI-related descriptions on a list

{
  "languages": ["python"],
  "keywords": ["agent", "inference", "model"],
  "minStarsToday": 20,
  "maxResults": 10
}

Keywords use literal, case-insensitive OR matching across repository names and descriptions. They are applied after fetching a current Trending page. An empty match is a valid outcome; this is not a GitHub-wide search.

Emit only new or changed observations

{
  "languages": ["python"],
  "maxResults": 25,
  "comparePrevious": true,
  "onlyChanges": true,
  "snapshotName": "python-discovery"
}

The first comparison run establishes a baseline and emits matching observations as new. Later runs can produce zero rows when nothing has changed. SUMMARY separates that result from failed source requests.

Input controls

Input	Default	Limits and behavior
languages	['python'] when empty or omitted	Up to 3 GitHub language slugs; case-insensitive deduplication; all selects the overview
keywords	[]	Up to 10 literal phrases, at most 100 characters each
minStarsToday	0	Minimum displayed growth; positive values exclude missing counts
maxResults	5	1–75 output rows across the run; languages are processed in input order
comparePrevious	false	Compare against a saved baseline in your account
onlyChanges	false	Enables comparison and suppresses unchanged rows
snapshotName	default	1–80 letters, digits, underscores or hyphens
maxRunSeconds	90	30–120 second application budget

Choose actual GitHub language slugs. An unsupported or currently empty list can fail validation rather than returning invented repositories. There are no arbitrary URL inputs and no weekly or monthly query modes: the current robots policy restricts those date-window query URLs.

Pricing and result limits

The configured result price is $0.002 per saved repository observation: 5 rows are $0.01 and 1,000 rows are $2 in result charges. The Actor's live Pricing tab is authoritative for any separate startup charge or platform fees.

Unchanged rows suppressed by onlyChanges, rejected pages and filtered-out repositories do not produce result events. Set Apify's maximum total charge as a second limit. The Actor checks the remaining charge allowance before output and stops at the platform's reported limit. A source outage may still incur applicable startup or platform charges.

Saved comparison behavior

Comparison history lives in the named key-value store github-trending-discovery-monitor-history in the account running the Actor. The language list, filters and snapshotName define its scope. A different scope starts separate history. Run only one writer per scope at a time; use separate snapshot names for independent workflows.

Each source page is parsed and validated completely before emitting any of its observations. History advances only for dataset output acknowledged as saved. A limited or partial run can update those confirmed rows, while the summary reports its limits. Previously seen repositories missing from today's list are not labeled removed or deleted. This Actor does not compare repository content or commits.

If an output request fails after delivery but before acknowledgement, the remote dataset may contain the row even though this run cannot confirm it. The Actor stops instead of retrying an uncertain charge, and does not advance that row's history. Downstream consumers can deduplicate by recordId, contentHash and observedAt.

Coverage, failures and responsible use

• These are current daily Trending lists, not every GitHub repository, every new project or an exhaustive popularity ranking.
• Trending presence, stars and rank changes do not establish adoption, revenue, quality or investment value.
• Source HTML and availability can change. HTTP errors, robot exclusions, challenges, malformed counts and unexpected page structures are surfaced; no cached or fabricated data is substituted.
• A transient HTTP 502/503/504 can receive one retry within the run budget. HTTP 429 stops remaining source work. No proxy or alternate access route is used to bypass a refusal.
• Output caps apply after filtering. If an earlier language fills the cap, later languages may not be reached. Read sourceCoverage, sourcesNotReached, selectionTruncated, incomplete and status in SUMMARY before treating a run as complete.
• The source's missing values remain null. Empty keyword matches can succeed with zero rows; unavailable or invalid source pages cannot.

Automation and delivery

Save an Apify task for a repeatable input, then schedule it at a cadence appropriate for a daily source. Send successful datasets to your existing spreadsheet, database, n8n or Make workflow through Apify integrations. For AI workflows, pass the dataset rows together with SUMMARY so the consumer can distinguish limited coverage from a complete selected check.

Use the Actor's API tab for the exact endpoint and input schema. Public source links and timestamps remain attached to each observation for downstream verification.

Need repository watchlists instead?

This Actor discovers projects from Trending lists. For a fixed repository watchlist, use our GitHub Repository Change Monitor. Browse Mayd It data tools for related workflows and examples.

For an issue, use the Apify Issues tab and include the run URL, selected language and expected behavior. Do not include account tokens or private credentials.