VouchVouch
Run a scan
Live · Built for the AI-coding era

Code, trusted before it's deployed.

Score GitHub repos before you trust them. Summarise audits, whitepapers, and specs in seconds. AI-driven analysis for engineering teams who can't afford to read every line themselves.

Scan a repoScan a documentHow it works
vouchprotocol.tech/scanlive preview
File matches · 14
  • src/auth/session.ts92%
  • src/api/router.ts78%
  • tests/setup.spec.ts47%
  • scripts/deploy.ts12%
  • lib/utils.ts8%
diff: src/auth/session.tsorigin → derivative
12 class Session 
13 { 
14   authenticate(...) { 
15     assert(...); 
16 }
12 class Session 
13 { 
14   authenticate(...) { 
15     assert(...); 
16 }
AI Analysis
38
caution

Mixed. 4 files match upstream open-source frameworks. Repo age 2 mo, single contributor.

derivative2 mo old

Trusted by engineering teams, OSS maintainers, and code auditors

Open-source maintainersCode reviewersSecurity researchersEngineering leadersIndependent auditors

Trust Layer for AI-built code

Scan a repo

Read every doc in seconds

Scan a document

Provenance API for engineering teams

Read the docs
Built for builders & agents

Trust before you deploy.

A code-trust score, a sub-second similarity diff, and an AI- written verdict — wired into the surfaces your traders, agents, and audit teams already use.

// The three signals

Provenance.Who wrote it. When. Why.
Similarity.Where else it lives.
Trust.Verdict, not vibes.
/01

Built for the AI-coding era

Engineering teams shouldn't just trust. They should verify, score, and audit — automatically and in seconds.

  • Score every commit, every contributor, every dependency.
  • Map upstream copies before you ship.
  • Cache scans across consumers — pay once, share forever.
/02

One Scan.
Full Picture.

A web scanner, REST API, and embeddable iframe powered by Claude — paste any github.com/owner/repo and Vouch returns a trust score, file diffs, and a written verdict.

  • · 0–100 trust score
  • · Side-by-side file diff
  • · Written origin verdict
Trust score
95/100
Lower Risk · Originator
A+
File matches5
  • src/auth/session.ts12%
  • src/api/router.ts8%
  • src/lib/utils.ts92%
  • scripts/deploy.ts4%
  • tests/auth.spec.ts47%
Derivative repos
  • Archiveopen-archives/sample-pack3
  • Auditaudit-references/sdk-toolbox2
  • Templatetemplates/api-starter1
/03

Code Audit for AI-Generated Repos

Vouch flags AI-generated copies, forks, CTF datasets and academic mirrors before you deploy or invest. Every file scored, every match traced.

  • AI-generated
    Distinguishes Claude/GPT-shaped repos from genuine human code.
  • Forks & rebrands
    Surfaces re-skinned copies even when files were renamed.
  • CTF / academic
    Tags datasets and challenge mirrors so they don't kill your score.
Introducing

Introducing RepoScan

GitHub Code Search unlocked discovery. For the first time, you can find exact copy-paste across millions of repos. But surfacing matches is only step one. Scoring trust is step two.

Designed to turn code review into infrastructure.

Read the spec
reposcan / livev1.0
vercel/next.js96
scan_8a3...e2foriginator · 12s
ai-gen/api-clone18
scan_4d1...8c7derivative · 9s
audit-tool/v371
scan_2b9...01amixed · 11s
anon/template-fork22
scan_f70...cdederivative · 7s
● live SSE3.2k scans / 24h

Designed to turn repos and documents into verdicts.

  1. .01

    Submit a repo or a doc

    Paste any github.com/owner/repo, or drop a PDF/DOCX/MD/TXT into the document scanner. Same shell, same speed — pick whichever matches the artifact in front of you.

  2. .02

    AI reads & searches

    For repos: pull source files, commit history, contributors, then run file-level similarity search across the public corpus. For documents: extract the text and run your preset (or your own prompt) through a focused analysis pass.

  3. .03

    Score, diff, verdict — or summary

    Repos return a 0–100 trust score, side-by-side diff, and a written origin-vs-derivative verdict. Documents return a streamed markdown answer — summary, key points, risks, action items, or whatever you asked for.

    4. Learn more

Why choose Vouch?

Vouch is building the trust layer for the AI-coding era — scanner, REST + SSE API, and embed iframe built around a single repo-provenance score.

AI-driven

Claude reads code in context — not regex. Catches refactored copies and re-skinned scams.

Real diff

Side-by-side viewer maps every match to its origin file with line-level highlighting.

Multi-source corpus

Indexes academic datasets, public archives, audit references and open-source production code.

Built for agents

Bearer-auth REST, sub-second SSE progress, sessionToken polling — autonomous-ready.

// By the numbers

Real scans. Real verdicts. Already in motion.

RepoScanlive
12,400+
Total scans
3,820
Unique repos
1.4M
Files indexed
Scans per day · last 30 days
REST APIlive
47k
API calls
184
Active keys
61
Embeds live
API calls per day · last 30 days
Questions

Frequently asked.

  • Vouch is an AI-powered code & document scanner for engineering teams. Paste a GitHub repo to score originality 0–100 and surface derivatives, or upload a PDF/DOCX/MD/TXT to get a tight summary, key points, risks, or any answer you want about it.

// Become one with Vouch

Trust before you deploy.
Verify before you ship.

Embed the iframe on your project page, wire the REST API into your audit pipeline, or open the scanner directly. Free tier covers casual use; strategic-collaboration spots are open.

Scan a repoStrategic CollaborationInvestor Relations
Avg. scan time
~14s
Per scan cost
$0.0001
Free tier
3 / hr
Auth
Bearer · IP