RepoPreflight
Preflight check for Git repositories

Know before you clone.

Surface execution risk, AI-agent risk and supply-chain signals in any public Git repo — before you clone it or hand it to your coding agent.

Public GitHub repos · Static analysis only · No cloning, no install, no execution

Example report

acme-labs/postgres-helper

Review recommended

Two install-time signals and an AI-agent instruction file modify shell behaviour.

Confidence: Mediumbased on 1y history · 4 contributors · 220 stars
Trust score
62
Open risk82
Install risk41
AI-agent risk55
Repo maturity68

Illustrative example. Higher scores reflect fewer observed risk signals — not a guarantee of safety.

Workspace & install risk

Inspects VS Code tasks, devcontainer, Dockerfile, and package install scripts.

AI-agent manipulation

Reads AGENTS.md, .cursorrules, Copilot instructions for prompt-injection patterns.

Web reputation

Looks up public reports, complaints and warnings about the repo before you trust it.

How we build the score

Sources we inspect

  • · GitHub repository metadata (age, contributors, releases, activity)
  • · Repository structure and lifecycle files
  • · Install scripts: package.json, setup.sh, Makefile, Dockerfile
  • · Workspace files: .vscode, devcontainer.json, GitHub Actions
  • · AI-agent instructions: AGENTS.md, .cursorrules, CLAUDE.md, Copilot
  • · Public reports and discussions about the repository

What we report

  • · Observed risk indicators with file + evidence
  • · Per-dimension scores (open, clone, install, AI-agent, maturity)
  • · A confidence level reflecting how much signal the repo provides
  • · Positive and negative signals, with rationale
  • · Recommended next steps — not absolute verdicts

We never claim a repository is "safe" or "malware-free". Reports describe what we observed in public data, with the confidence we can attach to it.

Plans

Free

$0
  • · 5 scans / day
  • · Public GitHub repos
  • · Basic trust report

Pro

$10 / mo
  • · More scans
  • · API access
  • · Scan history
  • · Advanced checks
  • · CLI & IDE plugin
  • · Browser extension

Enterprise

Talk to us
  • · Everything in Pro
  • · Team license
  • · Private repos
  • · SSO
  • · Audit logs
  • · Custom deployment

API

Programmatic access — currently open, rate-limited by IP.

POST /api/analyze
Content-Type: application/json

{
  "repository_url": "https://github.com/owner/repo"
}

CLI & IDE plugin

scangit + rp-clone wrapper, plus Cursor rule and Claude Code skill that force a preflight before cloning.

  • · scangit + rp-clone CLI
  • · .cursor/rules
  • · .claude/skills

to unlock.

Browser extension

Trust badge next to every GitHub repository name. Click for the full report. Chrome / Edge / Brave / Arc.

  • · Manifest V3
  • · Inline badge on github.com
  • · Configurable endpoint

to unlock.

Slack slash command

/preflight <url> in any channel. Posts the trust report back inline so the team can review before cloning.

  • · Signed requests
  • · Async response
  • · Per-workspace install
POST /api/public/slack/preflight