All projects

#projects · Developer security

Safeship.

A security co-pilot for people who build with AI tools. Connect a GitHub repo and it finds leaked secrets, insecure code, and vulnerable dependencies, then explains each fix in plain English.

Year
2026
Platforms
Web
Users
20+
Type
Web + AI
Safeship Advisor page for reviewing a repository's schema, tech stack, and structure

Overview

Safeship is a security co-pilot for people who build with AI tools. Connect a GitHub repository and it scans the code for leaked secrets, insecure patterns, and vulnerable dependencies, then explains every finding in plain English, ranked by real-world risk, with a copy-paste fix.

Scans run on on-demand GitHub Actions runners with gitleaks, Semgrep, and OSV-Scanner. Secrets are redacted before anything leaves the runner, and the cloned code is destroyed when the job ends. It is static analysis only: Safeship never probes or sends traffic to live systems.

Key features

  • Secret, code (SAST), and dependency scanning in a single run
  • Plain-English explanations and a 0–100 safety score for every scan
  • Fix with AI: opens a pull request that resolves a finding
  • Advisor reviews a repo's schema, tech stack, and structure, and draws its tables
  • Assistant chat for security and code questions
  • Runs on Groq in the cloud or fully local with Ollama

Built with

  • Next.js
  • Semgrep
  • Groq
  • GitHub Actions
  • gitleaks
  • OSV-Scanner
  • Prisma
  • PostgreSQL
  • Auth.js
  • Ollama