Social Intelligence
GitHub Activity Signals
Published Updated
Portfolio-level engineering signals from public GitHub activity: AI/ML investment, developer tooling, infrastructure scaling, security focus, and more. Roughly 250,000 signals a month across 86,000+ companies.
- Companies Covered
- 15M+
- Signal Subtypes
- 8
- Metrics Tracked
- Stars, Forks, Languages
- Delivery
- REST API, GCS Push, Generate Insights API, Flat File
- Format
- JSONL or Parquet
- Bucket
- gs://autobound-github-v1/
What Are GitHub Activity Signals?
Monthly refresh
GitHub activity signals track what a company's engineering team is building, measured across its entire public repository portfolio rather than individual repos. Autobound maps GitHub organizations to company domains for 15M+ companies and fires a signal when activity crosses meaningful thresholds: star growth above 20% in 90 days, a cluster of new AI/ML repositories, or sustained investment in infrastructure and security tooling.
Subtypes cover AI/ML investment, developer tooling, open-source investment, infrastructure scaling, security focus, and new project launches. Each signal carries portfolio metrics: repository count, star and fork growth over 30, 60, and 180 days, top repositories, and the technologies involved. At roughly 250,000 signals a month across 86,000+ companies, this is observable engineering behavior, visible before any buying process starts.
Example subtypes
AI/ML investmentgithubAIMLInvestment
Illustrative example
Vercelvercel.com
Vercel's AI SDK repo grows 34% in 180 days as the company deepens its AI tooling investment
Source: vercel.com
From a GitHub repo to your next move
What we detect, what you get, and what you do with it. Pick the path that fits your team.
Sales, marketing and RevOps: Slack, CRM fields and plays.
Sales, marketing and RevOps: Slack, CRM fields and plays.
0103
Engineering teams build in public
We watch each company's public GitHub portfolio and fire a signal when activity crosses a threshold, such as fast star growth or new AI repos.
- GitHub
- We detectGitHub
- You getSignal record
- You actSlack, CRM or agent
GitHubJul 14, 2026The ai repository (Vercel AI SDK) added 3,400 stars over the past 90 days, driven by adoption of streaming and agent patterns in production apps. Combined with two new inference-related repos, this signals sustained AI/ML investment across the portfolio...
ExtractedVercelvercel.comcompanyAI/ML investmentsignal_subtype0203
You get one clean record
Each record names the company and its domain, labels what happened, and links back to the source.
- Signal record
- We detectGitHub
- You getSignal record
- You actSlack, CRM or agent
Signal recordVercel
vercel.comDeveloper Tools
AI/ML investmentgithubAIMLInvestmentVercel's AI SDK repo grows 34% in 180 days as the company deepens its AI tooling investment
“The ai repository (Vercel AI SDK) added 3,400 stars over the past 90 days, driven by adoption of streaming and agent patterns in production apps. Combined with two new inference-related repos, this signals sustained AI/ML investment across the portfolio...”
- Repos
- ai, next.js
- Detected
- Jul 14, 2026
- Source
- github.com
signal_id c9d34e56…
0303
Act on it where you work
Send it to Slack, write it to your CRM, or give it to an AI agent.
- When
signal_subtype= githubAIMLInvestment- Do
- Dev-tool outreach to the engineering lead that names the repo
- Your team acts
- We detectGitHub
- You getSignal record
- You actSlack, CRM or agent
or your own code#account-signalsSignal alert from your workflow
Vercel· AI/ML investment
Vercel's AI SDK repo grows 34% in 180 days as the company deepens its AI tooling investment
Play: Dev-tool outreach to the engineering lead that names the repo
Start the playOpen the sourceIllustration. The record fields are real.
0103
Engineering teams build in public
We watch each company's public GitHub portfolio and fire a signal when activity crosses a threshold, such as fast star growth or new AI repos.
0203
It lands in your stack
Land weekly JSONL from your GCS bucket. Join on
company.domain; LinkedIn URL is almost never present.0303
Ship it in your product
Add
technologies_mentionedas a technographic filter.
- GitHub
- We detectGitHub
- You getSignal record
- You actSlack, CRM or agent
The ai repository (Vercel AI SDK) added 3,400 stars over the past 90 days, driven by adoption of streaming and agent patterns in production apps. Combined with two new inference-related repos, this signals sustained AI/ML investment across the portfolio...
Step 1 of 3: Engineering teams build in public
Try it on your own accounts
1,000 free credits. Query github-initiative in the Signal API or from Claude.
How we build this signal
How we build each GitHub signal
Every month we re-measure the public repositories of each tracked company and measure real growth against our stored snapshot history.
- Companies tracked
- ~89,000
- Repositories measured
- ~187,000
- Growth windows
- 30 / 60 / 180 days
- Refresh
- Monthly
- 01
Repository universe
We track about 89,000 companies and their top 3 public repositories, about 187,000 repositories in all.
- 02
Current stats and READMEs
Each run reads every repository's current stars, forks and watchers. READMEs are read only for repositories pushed in the last 60 days, so dormant code does not drive a signal.
- 03
Measured growth
Every run stores a snapshot, and growth is measured against snapshots from about 30, 60 and 180 days earlier. A repository counts in a growth figure only when it has a baseline for that window; with no baseline, growth is 0, never estimated.
- 04
Extraction
A model reads the stats, growth and READMEs and must pick from 8 fixed subtypes, such as new project launch, AI and ML investment or infrastructure scaling. It may not claim growth where measured growth is 0, or infer it from star counts.
- 05
Dedup and delivery
Each signal is keyed on company domain, subtype and the repositories it names, so the same finding never ships twice.
GitHub Signal Data Schema
GitHub signals include portfolio-level metrics, repository trends, technology categorization, and growth velocity measurements.
{
"signal_id": "c9d34e56-7b2a-4f19-8c83-1e5d9f0a2b67",
"signal_type": "github-initiative",
"signal_subtype": "githubAIMLInvestment",
"detected_at": "2026-07-14T09:22:31Z",
"batch_id": "gh-20260714-89aa726e-c725-454d-8afc-4d5fb1e1984c",
"association": "company",
"company": {
"name": "Vercel",
"domain": "vercel.com", // match on domain
"linkedin_url": "linkedin.com/company/vercel", // or match on LinkedIn URL
"industries": ["Developer Tools", "Cloud Infrastructure"],
"description": "Frontend cloud platform for AI-native web applications..."
},
"contact": [],
"data": {
"summary": "Vercel's AI SDK repo grows 34% in 180 days as the company deepens its AI tooling investment",
"detail": "The ai repository (Vercel AI SDK) added 3,400 stars over the past 90 days, driven by adoption of streaming and agent patterns in production apps. Combined with two new inference-related repos, this signals sustained AI/ML investment across the portfolio...",
"relevance": 0.87,
"confidence": "high",
"sentiment": "positive",
"referenced_repos": ["ai", "next.js"],
"technologies_mentioned": ["TypeScript", "LLM streaming", "React Server Components"],
"portfolio_metrics": {
"repository_count": 142,
"growth": {
"stars_pct": { "30d": 0.06, "60d": 0.14, "180d": 0.31 },
"forks_pct": { "30d": 0.05, "60d": 0.11, "180d": 0.24 }
},
"velocity": { "avg_stars_per_repo_30d": 41.7 },
"concentration": { "top_3_star_share": 0.52 }
},
"top_repositories": [
{
"name": "ai",
"full_name": "vercel/ai",
"url": "https://github.com/vercel/ai",
"description": "The AI Toolkit for TypeScript, for building AI-powered applications",
"first_seen_at": "2023-05-25T17:13:41Z",
"current": { "stars": 16800, "forks": 2900, "watchers": 16800 },
"growth_pct": {
"stars": { "30d": 0.09, "60d": 0.19, "180d": 0.34 },
"forks": { "30d": 0.07, "60d": 0.15, "180d": 0.28 }
}
}
]
}
}Bulk data
Need this signal for 10,000+ companies?
Get a free sample file or a bulk feed (API, GCS push, flat file). The sample is real GitHub Initiatives records, as a CSV.
Measured coverage
What's inside GitHub Activity Signals
A dated snapshot counted in our production database, not a live count. Minimums, not ceilings.
Top subtypes
8 totalGeography
all 134 countries →Top industries
all industries →Join keys
fill rateSee for yourself
Check out sample data: real GitHub Activity records with the same fields the API returns.
“Autobound's API streamlined our data management, accelerated product development, and improved outreach quality.”