GitHub

View API reference · Tag GitHub

We built GitHub matching so a username (a contributor, a stargazer, a username on a résumé) becomes a LinkedIn profile URL, and optionally the work emails attached to that developer's commit history. It also runs the other way: give it a LinkedIn URL or a name and it finds the GitHub account.

Developers are the hardest audience to reach from a conventional B2B dataset. Their LinkedIn is often stale and their work email is rarely listed, but their code is public and current. This is the bridge between the two.

Once you have a LinkedIn URL, the person is an ordinary Fiber profile. Email and phone beyond what GitHub exposes come from Contact Reveal.

Which direction do you need

Anything described as a background job follows the same shape: you call the trigger, it returns a run ID, and you either poll for results or subscribe to the completion webhook. The first poll is normally still running — that is expected, not a failure.

GitHub → LinkedIn

What you send

What you get back

Single request — POST /v1/github-to-linkedin/single:

{
  "apiKey": "YOUR_API_KEY",
  "githubUsername": "torvalds",
  "outputType": "both",
  "context": "maintainers of large open-source infrastructure projects"
}

An unknown username returns 404, and so do bot accounts and organization accounts — the route resolves people, not orgs. Those calls are refunded in full. A username that exists but has no confident LinkedIn match returns 200 with linkedInUrl null, and you are billed only for the parts that actually resolved.

Polling the batch

The batch poll takes the githubAgentRunId from the trigger. Page through results with cursor, sending back the nextCursor from the previous response; pageSize runs 1–100 and defaults to 10. The run-level status moves NOT_STARTED → STARTED → DONE (or FAILED), and each row carries its own status. You have everything when the run is DONE and hasMore is false. The github_to_linkedin.completed webhook fires at the same point.

How long a batch takes grows with the size of the list and whether you asked for emails, so don't hold a connection open waiting for it. Take the webhook, or poll about once a minute.

LinkedIn or a name → GitHub

This direction is a background job only.

The LinkedIn → GitHub trigger takes people[] (up to 1,000) and, unlike the other routes, requires overallContext: one sentence describing what the list has in common, such as "backend engineers at Series B fintechs in NYC". Each person is identified one of three ways:

• inputType: "linkedinUrl" — a LinkedIn URL, slug, or entity URN
• inputType: "linkedinUserId" — the numeric LinkedIn user ID
• inputType: "manual" — fullName plus optional company, jobTitle, schoolName, workEmail, and linkedinUrl

Pass externalId on each row to join the results back to your data.

Its poll takes the githubAgentRunId and returns the whole result set in one response; there is no paging on this route. You get:

Treat counts.pending reaching 0 as the finish line, or subscribe to github_lookup.completed and skip polling entirely.

Using it effectively

Use the batch route for lists. It costs the same per person as the single route (5 credits for a LinkedIn match, 3 for email), but one call covers up to 1,000 people and runs in the background, so you don't spend your rate limit on a loop. Anything that is genuinely a list (a repo's contributors, a stargazer export, a conference attendee sheet) belongs on the trigger.

Ask only for the output you will use. outputType: "both" on the single route costs 5 credits for the LinkedIn match plus 3 more for email extraction. If you are going to run Contact Reveal anyway, linkedin alone is the cheaper path to the same place.

Send context, especially on batches. overallContext is required on the LinkedIn → GitHub direction for a reason: matching a name to a GitHub account is much easier when the system knows the list is all Rust compiler contributors rather than all people called Chen. The optional context on the GitHub → LinkedIn side pays off the same way.

Use confidenceOutOf10 as a routing decision, not decoration. Pick a threshold, send everything above it straight into outreach, and put the middle band in front of a human. A null linkedInUrl and a 3-out-of-10 match are very different problems.

Filter developers by what they build. The githubProfile on a result gives you company, location, and bio as the developer maintains them, which tends to be fresher than a self-reported LinkedIn headline for this audience.

Prefer the webhook to a poll loop. Both directions fire a completion event. Polling is free but it is still code you have to write, back off, and monitor.

Use cases

Turn a competitor's contributor list into a recruiting pipeline. Export the usernames who have committed to a relevant open-source project, send them through the GitHub → LinkedIn batch route in one call with outputType: "linkedin" and an overallContext describing the project, take the github_to_linkedin.completed webhook, then run Contact Reveal on the LinkedIn URLs above your confidence threshold. The batch charges 5 credits per person when it starts, so filter the list to the people you'd actually contact before you run it.

Reach developers already using an adjacent technology. Stargazers of a complementary library are the warmest developer-tool audience there is. Batch-resolve them the same way, then segment on the company field from githubProfile so you are not pitching to people at your own customers.

Qualify one candidate mid-conversation. A recruiter looking at a résumé with a GitHub username on it wants an answer now, not a job. That is what the single lookup is for: one call, outputType: "both", LinkedIn and any commit emails in the response.

Find the GitHub account behind a LinkedIn list. When your list starts on LinkedIn (a saved search, a target account's engineering team), run the LinkedIn → GitHub trigger with a tight overallContext and read counts.found to see how technical the list really is. The GitHub URLs give you something concrete to reference in a first message.

Credits

Costs depend on direction and on whether you batch:

Batch triggers charge when the job starts, which shows up as charged-for-async-process. Rate limits differ per route: the single lookup allows 120 requests per minute, the triggers 30, and the GitHub → LinkedIn poll 60. chargeInfo on the response is the authoritative amount — see Billing.

Credit amounts are defaults — custom pricing is available for bulk use cases.

Related: Social media APIs · X → LinkedIn · Contact Reveal · Webhooks