0–19
Curious
No validated wins in our benchmark — yet. Either a fresh GitHub account, a star history that predates the modern OSS-VC wave, or a taste profile that hasn't aligned with venture-stage outcomes so far. Everyone starts here.
Loading...
The proprietary metric
Scout Score is a 0–100 measure of investment taste computed from a developer's public GitHub starring history — specifically, how many validated unicorns, big-funding, and acquisition events they starred before the event happened. The same way Ahrefs owns DR as the Code-Side Sourcing category's authority metric, VC Deal Flow Signal owns Scout Score as the backwards-looking taste metric for developer-investors. Free, public, reproducible from the GitHub stars API.
Try it free
The live tool lives at /receipts. Paste any GitHub username, get the score, the rank, the top 8 early calls, and a shareable card. This page is the canonical definition; the tool is where you run the calculation.
Scout Score is a proprietary metric defined by VC Deal Flow Signal (GitDealFlow) that grades a GitHub user's backwards-looking investment taste from their public starring activity. The input is the user's starred-repo timeline; the benchmark is a curated panel of ~75 validated unicorns, large funding rounds, and acquisitions. The output is a single 0–100 score plus a ranked list of the user's earliest correct calls.
The metric is defined by four properties: (1) input data is fully public and reproducible from the GitHub stars API, (2) the benchmark panel is published and versioned, (3) the scoring algorithm is open-source and deterministic — no AI judgement, no proprietary weighting, (4) the score is backwards-looking only. Scout Score measures whether you *saw it coming*, not whether you will see the next one coming.
Scout Score is not a credit score, a reputation score, or a contributor score. It measures one narrow thing: taste in identifying venture-stage companies before the market did, as evidenced by the only public timestamped trail most developers leave — their GitHub stars.
Every Scout Score maps to one of five named tiers. The tiers are structural, not cohort-adjusted — the same thresholds apply to every user regardless of account age or star volume.
No validated wins in our benchmark — yet. Either a fresh GitHub account, a star history that predates the modern OSS-VC wave, or a taste profile that hasn't aligned with venture-stage outcomes so far. Everyone starts here.
At least one early call landed. The user starred a company that went on to a unicorn round, acquisition, or $1B+ valuation, and they starred it before the event. A real but thin track record.
Multiple early calls across two or more orgs. The pattern is no longer luck. This is the threshold where the score starts to be useful as a signal of repeatable taste.
A consistent record of starring winners before they were obvious. At this tier the starred-repos feed is itself a leading indicator — high-Elite scouts tend to surface the next wave 6–12 months before the broader market notices.
Near-perfect taste across five or more validated wins, with significant lead time on most. The top of the ladder. An Oracle scout's new stars are worth tracking as a sourcing channel in their own right.
Five deterministic steps. No AI judgement, no proprietary weighting, no curve. The algorithm is published here in full — anyone can reproduce the number from the public GitHub stars API and the published validated-wins panel.
Via the GitHub stars API — a fully public endpoint that returns every public repo the user has starred with a timestamp. No authentication on the user's side, no private data read, no scopes granted. The data was always public; Scout Score just grades it.
We maintain a curated, versioned list of ~75 companies that hit a venture-stage event (unicorn valuation, large primary round, acquisition) between 2020 and now. Each entry carries an event date and a weight (25–100) reflecting the event's significance — a $100M Series D weights higher than a $10M Series A.
For each star that lands on a validated-win repo, we compute months between the star timestamp and the event date. If the star predates the event, it's an early call and earns points. If it postdates, it's a late star — counted as a match but worth zero points. Half the signal is the timing, not the pick.
A user can star multiple repos from the same company (the monorepo, the SDK, the docs site). We keep only the earliest star per organisation so one company can't inflate the score. The top 5 wins by points form the headline number.
Five perfect wins (5 × 100 max-weight points) normalise to score 100. The formula is transparent: top5_points_sum ÷ 500 × 100, capped at 100. No curve, no cohort adjustment, no hidden prior. Anyone can reproduce the number from the public star timeline and the published panel.
A high Scout Score does not mean the user is influential, respected, or followed. It means their starring history aligned with venture-stage outcomes. A quiet developer with 50 followers can outscore a celebrity with 50,000.
Scout Score ignores commits, PRs, and issues entirely. It reads only stars — the lightest-weight signal GitHub exposes. The metric measures taste in identifying code worth bookmarking, not engineering output.
Scout Score is strictly backwards-looking. A high score says 'this person saw the last wave early' — it does not say they will see the next wave. The forward-looking product is the weekly Acceleration Watch, not the Scout Score.
It has no bearing on hiring, lending, insurance, or any decision outside venture sourcing. The only legitimate use is as a sourcing input: high-scout feeds are worth watching because their new stars correlate with later venture-stage events.
Yes. The tool at /receipts is free, requires no login, no email, and no GitHub OAuth scope beyond public read. Paste a username, get a score in under 10 seconds. The MCP tool get_scout_receipts is also free and never gated.
No. Scout Score reads only your public starred-repo timeline — the same data anyone can see by visiting your GitHub profile and clicking the Stars tab. No private repos, no commit content, no email, no OAuth scopes beyond public read.
The median active developer scores in the 5–25 range — most stars predate the modern OSS-VC wave or land on repos outside the validated panel. A score of 40+ (Sharp tier) is genuinely rare and indicates repeatable early-call behaviour. Scores above 60 (Elite) are uncommon enough that the starred-repos feed itself becomes a sourcing signal worth tracking.
DR measures the backlink-based authority of a domain. Scout Score measures the starring-taste of a GitHub user. DR is about citation weight; Scout Score is about call timing. The two metrics operate on completely different surfaces (web link graph vs GitHub star graph) and answer different questions. The structural similarity is that both are proprietary 0–100 metrics owned by a single vendor that publishes the methodology — DR is to backlinks what Scout Score is to GitHub stars.
Not by gaming it. Starring more repos won't help — the score rewards early calls on the right repos, not star volume. The only honest way to raise a Scout Score is to develop taste for engineering-led companies before their breakout, which is the skill the metric is measuring. The cleanest side benefit: if you star a company early and it later validates, your score goes up retroactively.
The panel is curated from Crunchbase funding announcements, acquisition records, and public valuation disclosures between 2020 and now. We publish the list of ~75 companies (with event dates and weights) as part of the open dataset. The curation is human-reviewed; additions and removals are versioned.
Best next step
The definition you just read is the canonical surface. The computation lives at /receipts. Paste your GitHub username, get the score in under 10 seconds.