Skillget

airtight-secure-coding

agent
Guvenlik Denetimi
Basarisiz
Health Uyari
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Basarisiz
  • process.env — Environment variable access in validation/with-airtight/config.js
  • exec() — Shell command execution in validation/with-airtight/db.js
  • fs module — File system access in validation/with-airtight/db.js
Permissions Gecti
  • Permissions — No dangerous permissions requested

Bu listing icin henuz AI raporu yok.

SUMMARY

Secure-coding gates for AI-written code — 67 gates across 13 OWASP-aligned topics, checked before code ships.

README.md

airtight — secure-coding gates for AI-written code

MIT works with Claude Code, Cursor, Codex 67 checks across 13 topics mapped to OWASP Top 10 + CWE GitHub stars LinkedIn

AI writes your code. Who checks it?

Ask for a login and you get a good one — argon2, httpOnly cookies, ownership inside the query. The model knows all that. What it skips is the part nobody asked for: the rate limit, the lockout, the audit log. Guess the password thirty times and nothing stops you — no throttle, no record, invisible.

Airtight is the part of the request nobody makes: hard gates on what the assistant may emit, checked before the code reaches you.

flowchart LR
  A[You ask for a feature] --> B[Model writes it]
  B --> C{Airtight<br/>67 gates}
  C -->|fails| D[Rewrite]
  D --> C
  C -->|passes| E[Code that<br/>already passed]

Measured, not claimed

Same brief, two agents — one with Airtight, one without. A blind third scored and ran both.

24 applicable gates → 24/24 with Airtight, 21/24 without.

The three misses: no rate limit, no lockout, no security log. Verified live — thirty guesses returned thirty 401s with zero throttle and zero records; the Airtight build returned 429 and wrote ten audit lines. The control was good (scrypt, timingSafeEqual, httpOnly, ownership) — the gaps were omissions, not bad crypto.

Both apps and the full comparison → validation/.

Install

npx skills add Zyoffsec/airtight-secure-coding

Or clone it into your skills folder (Claude Code, Cursor, Codex):

git clone https://github.com/Zyoffsec/airtight-secure-coding ~/.claude/skills/airtight

Then forget it. Airtight runs itself — write code as usual and the gates apply silently. No command to invoke.

Usage

Command What it does
(default) You write code; gates apply silently.
airtight audit <target> Scores code against the gates. Read-only.
airtight harden <target> Finds and fixes gate failures.
airtight prove <target> Probes your local code with edge-case input.

What it catches

67 gates across 13 topics, mapped to OWASP Top 10 (2021) + CWE. Each is binary and numbered — every finding cites its gate. Full list → references/gates.md.

What it doesn't: business-logic bugs, unknown CVEs in your dependencies, or architecture review — those need a human. Airtight is the first check, on the mistakes mechanical enough to check mechanically. It never calls code "secure" — only which gates held.

Demo

A 30-second recording is coming — a good login, thirty silent break-in attempts, then the same with Airtight → 429 and a full audit log. It is the point of the whole project. Contributions welcome.

MIT License · Contributing

Yorumlar (0)

Sonuc bulunamadi