Pricing

Rating

Developer

Actor stats

Categories

Share

GitHub Profile Scraper — Users, Repos, Stars & Language Stats

Pull GitHub user profiles plus their public repositories — bio, follower counts, top repos by stars, language distribution, license info, activity timestamps — using the official GitHub REST API. No token required for public-data lookups (60 req/h unauthenticated rate-limit applies; the actor handles that automatically).

What you get per user

The actor pushes one record per username. All fields below come from GET /users/{username} and GET /users/{username}/repos — verified against src/main.js.

Profile-level fields (19, including scrapedAt metadata)

Repo-level fields (15 per repo, when includeRepos=true)

name, fullName, description, url, stars, forks, watchers, language, topics[], isForked, createdAt, updatedAt, pushedAt, license, openIssues

Plus aggregates on the parent profile record (computed from the EXTRACTED repos only — see Honest limitations):

• totalStars — sum across the extracted repos (capped by maxReposPerUser, NOT total stars across all of the user's repos)
• totalForks — sum across the extracted repos (same caveat)
• languages — array of {language, repoCount}, sorted desc by repo count. repoCount is "how many of the extracted repos have this as their language field" — this is NOT GitHub's bytes-weighted language graph; for that you need the /repos/{owner}/{repo}/languages endpoint per-repo (custom build available).

Output example

{
"username":"torvalds",
"name":"Linus Torvalds",
"bio":null,
"company":"Linux Foundation",
"location":"Portland, OR",
"email":null,
"blog":"",
"twitterUsername":null,
"avatar":"https://avatars.githubusercontent.com/u/1024025?v=4",
"profileUrl":"https://github.com/torvalds",
"publicRepos":7,
"publicGists":0,
"followers":220000,
"following":0,
"createdAt":"2011-09-03T15:26:22Z",
"updatedAt":"2026-04-20T10:11:12Z",
"hireable":null,
"type":"User",
"totalStars":185000,
"totalForks":55000,
"languages":[
{"language":"C","repoCount":4},
{"language":"Shell","repoCount":1}
],
"repos":[
{
"name":"linux",
"fullName":"torvalds/linux",
"description":"Linux kernel source tree",
"url":"https://github.com/torvalds/linux",
"stars":175000,
"forks":53000,
"watchers":175000,
"language":"C",
"topics":["linux","kernel"],
"isForked":false,
"createdAt":"2011-09-04T22:48:12Z",
"updatedAt":"2026-04-29T08:00:00Z",
"pushedAt":"2026-04-29T07:55:00Z",
"license":"GPL-2.0",
"openIssues":1
}
],
"scrapedAt":"2026-04-29T12:00:00.000Z"
}

Input

Use cases

• Developer recruiting — evaluate candidates by their open-source footprint, language mix, and active-repo cadence (pushedAt).
• Competitor analysis — fingerprint the OSS strategy of a company by mapping its top contributors' top repos.
• Community research — identify influential developers in a technology ecosystem (high stars + matching topics).
• Portfolio benchmarking — compare repo activity, fork ratios, and license distributions across a candidate set.
• Lead generation — surface developers with specific technology expertise (filter by languages[*].language).

How it works

1. GET /users/{username} → 18 fields from GitHub + scrapedAt (19 total).
2. (If includeRepos) GET /users/{username}/repos?sort=stars&direction=desc&per_page=<maxReposPerUser> → 15 fields per repo. Single page only — no pagination beyond GitHub's per-page ceiling (100).
3. Aggregate totalStars, totalForks, and languages from the extracted repo set (capped by maxReposPerUser).
4. On HTTP 403 with X-RateLimit-Remaining: 0, the actor reads X-RateLimit-Reset, sleeps until the window resets (+5 s buffer), then retries. No manual handling required, but unauthenticated mode caps you at 60 req/h — set up an authenticated proxy actor (custom build) if you need higher throughput.

500 ms delay between user-profile fetch and repo-list fetch; 1 s delay between users.

Honest limitations (read before bulk runs)

• No pagination beyond maxReposPerUser (GitHub per_page ceiling = 100). For users with hundreds of repos, only the top-N by stars are returned. If you need ALL repos, request a paginated custom build.
• totalStars / totalForks are EXTRACTED-repo aggregates, NOT lifetime totals. A user with 200 repos and maxReposPerUser=30 will show totalStars summed over the top 30 only — not all 200.
• languages is repoCount, NOT bytes-weighted. GitHub's UI shows percent of repo bytes per language; this actor counts how many repos list each language as their language field. Bytes-based language stats require per-repo /languages calls (custom build).
• One user's HTTP error halts the entire batch. The for (const username of usernames) loop is wrapped in ONE outer try/catch; a 502 / 503 on user #5 of 100 stops the run. Workaround: split large batches into ≤25-username runs, or request a per-user try/catch custom build.
• Rate-limit retry has no max-wait safeguard. If X-RateLimit-Reset shows 60 minutes ahead, the actor sleeps 60 minutes (+5 s) before retrying. For a 100-user batch hitting the unauthenticated cap, total wall-clock can stretch to 1.5–2 h. For high-volume runs, request an authenticated custom build (5000 req/h with PAT).
• No proxy. Direct fetch from the Apify worker IP. GitHub's unauthenticated rate-limit is per-IP, so co-tenant noise on the same Apify IP can shrink your effective quota.
• email is null for most users. GitHub only exposes email when the user has explicitly set a public email in profile settings.
• license may be null for repos without a declared license, or repos using a license GitHub can't classify (returns spdx_id only, or null).
• hireable is tri-state (true / false / null — GitHub-account opt-in). README example shows null (the most common state).
• topics ordering is GitHub-internal — not alphabetical, not by relevance; just the array as returned.
• Hardcoded UA ApifyGitHubProfileScraper/1.0. GitHub recommends a UA but doesn't inspect it for capability gating.
• Empty usernames = [] is silently accepted — actor logs No usernames provided. and exits 0.

Python integration

from apify_client import ApifyClient
client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("knotless_cadence/github-profile-scraper").call(
 run_input={
"usernames":["torvalds","gaearon","sindresorhus"],
"maxReposPerUser":20,
}
)
for p in client.dataset(run["defaultDatasetId"]).iterate_items():
 top_lang =(p.get("languages")or[{}])[0].get("language","?")
print(f"{p['username']:>20}{p['followers']:>7,} followers {p.get('totalStars',0):>7,} stars top:{top_lang}")

GitHub data toolkit (related actors)

All free to inspect on Apify Store — 31 published actors, 78 total in portfolio.

Common questions

Q: Do I need a GitHub API token? A: For public profiles, no — the actor uses the unauthenticated REST API (60 req/h). With ~30 repos per user, that supports about 30 users per hour before the rate-limit kicks in. The actor handles X-RateLimit-Reset automatically; very large batches will simply pause and resume.

Q: Does this scrape stars given, contribution graph, or sponsors? A: No — those are separate REST endpoints (and the contribution graph is HTML-only). Available as a custom build (see Custom scraping below).

Q: Why is email null even on public-figure accounts? A: GitHub only exposes email when the user has explicitly set a public email in profile settings. For most profiles email is null. Use the blog field as a fallback contact-discovery signal.

Q: Are forked repos included? A: Yes — isForked: true flags them. Filter them out client-side if you want owned-only repos.

Custom scraping — pilot tiers

Need authenticated batches, contribution-graph extraction, organization-wide audits, or a different schema (e.g. last-90-days commits per repo)? Three tiers:

• Pilot — $97 · 1 actor, basic config, 7-day support. Good entry point — useful for a single OSS-strategy report on one company's contributors.
• Standard — $297 · custom actor + Slack/email alerts on results, 30-day support. Most recruiting and competitor-OSS projects fit here.
• Premium — $797 · custom actor + dashboard + 90-day support + 1 modification round. For ongoing pipelines (weekly contributor refresh, multi-org rollups, technology-trend tracking).

Email: spinov001@gmail.com — drop the username list and the schema you need; quote within 48h.

Proof of work: 31 published Apify scrapers (78 total in portfolio) — Trustpilot 949 runs, Reddit 80+, Google News 43, Glassdoor 37, Email Extractor 36+. Recently delivered a paid 3-article series for a client in the proxy industry ($150).

More tips: t.me/scraping_ai · blog.spinov.online

Disclaimer

Designed for recruiting research, OSS-strategy analysis, and academic use. Respect GitHub's Terms of Service, applicable data-protection law (GDPR, CCPA), and scrape publicly visible content only. Not affiliated with GitHub, Inc. or Microsoft Corporation.

Honest disclosure: 19 profile fields (18 from GitHub + scrapedAt) + 15 repo fields per record. Unauthenticated rate-limit is 60 req/h; the actor handles X-RateLimit-Reset automatically (no max-wait safeguard — long resets = long waits). totalStars/totalForks/languages aggregates are computed over the EXTRACTED repos only (capped by maxReposPerUser, max 100), NOT lifetime totals. languages.repoCount is repo count not bytes-weighted. Single API page, no pagination beyond 100 repos. One user's HTTP error halts the batch (outer try/catch). No contribution graph, no commit-activity series, no stars-given list, no sponsors data — those are different endpoints and can be built as custom additions.