Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
A SAST skill that gives AI coding agents structured vulnerability detection across 34 vulnerability classes.
| Date | Stars |
|---|---|
| 2026-07-31 | 273 |
| 2026-08-06 | 273 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# llm-sast-scanner
A general-purpose **Static Application Security Testing (SAST) skill** for LLM-based code vulnerability analysis. Designed to be loaded by AI coding agents (Claude Code, OpenAI Codex, etc.) to perform structured source-to-sink taint analysis across 34 vulnerability classes.
---
## What It Does
This skill gives an LLM agent a structured, evidence-based workflow for finding security vulnerabilities in source code:
1. **Load** relevant vulnerability reference files for the target codebase
2. **Map sources** — identify all entry points where attacker-controlled data enters
3. **Trace taint** — follow data flow through transformations to potential sinks
4. **Verify findings** — apply a Judge step to eliminate false positives
5. **Report** — produce actionable findings with file path, line number, and remediation
Supports **Java, Python, JavaScript/TypeScript, PHP, .NET** with language-specific detection rules.
---
## Installation
### Git (recommended)
```bash
# Claude Code
git clone https://github.com/anthropic-lab/llm-sast-scanner.git
cp -r llm-sast-scanner/llm-sast-scanner/ ~/.claude/skills/
# OpenAI Codex
git clone https://github.com/anthropic-lab/llm-sast-scanner.git
cp -r llm-sast-scanner/llm-sast-scanner/ ~/.codex/skills/
```
### Manual
Download and copy the `llm-sast-scanner/` directory into your skills folder:
```bash
# Claude Code
cp -r llm-sast-scanner/ ~/.claude/skills/
# OpenAI Codex
cp -r llm-sast-scanner/ ~/.codex/skills/
```
---
## Structure
```
llm-sast-scanner/ ← repo root
├── README.md
└── llm-sast-scanner/ ← skill directory (copy this)
├── SKILL.md # 6-step workflow + Judge verification
└── references/ # 34 vulnerability knowledge bases
├── xss.md
├── sql_injection.md
├── path_traversal_lfi_rfi.md
└── ... (34 files total)
```
### SKILL.md
The main entry point. Defines the detection workflow, taint propagation rules, and Judge verification protocol.
---
## Advanced Usage Tips
- **Precompute call graph before scanning** — improves cross-function reasoning and reduces missed paths
- **Run 2+ scanning rounds** — increases recall and stabilizes findings via iterative refinement
- **Enforce per-finding validation** — significantly reduces false positives through explicit verification
---
## Vulnerability Coverage
34 reference files covering the following categories:
### Injection
| File | Vulnerability |
|------|--------------|
| `sql_injection.md` | SQL Injection (CWE-89) |
| `xss.md` | Cross-Site Scripting (CWE-79) |
| `ssti.md` | Server-Side Template Injection |
| `nosql_injection.md` | NoSQL Injection |
| `graphql_injection.md` | GraphQL Injection / Introspection Abuse |
| `xxe.md` | XML External Entity (CWE-611) |
| `rce.md` | Remote Code Execution / Command Injection |
| `expression_language_injection.md` | Expression Language Injection (SpEL, OGNL) |
### Access Control & Auth
| File | Vulnerability |
|------|--------------|
| `idor.md` | Insecure Direct Object Reference |
| `privilege_escalation.md` | Privilege Escalation |
| `authentication_jwt.md` | JWT Vulnerabilities (alg:none, weak secret) |
| `default_credentials.md` | Hardcoded / Default Credentials |
| `brute_force.md` | Brute Force / Missing Rate Limiting |
| `business_logic.md` | Business Logic Flaws |
| `http_method_tamper.md` | HTTP Method Tampering |
| `verification_code_abuse.md` | Verification Code Abuse |
| `session_fixation.md` | Session Fixation (CWE-384) |
### Data Exposure & Crypto
| File | Vulnerability |
|------|--------------|
| `weak_crypto_hash.md` | Weak Cryptography (CWE-327), Weak Hash (CWE-328), Weak Random (CWE-330) |
| `information_disclosure.md` | Sensitive Information Disclosure |
| `insecure_cookie.md` | Insecure Cookie Flags (CWE-614, CWE-1004) |
| `trust_boundary.md` | Trust Boundary Violation (CWE-501) |
### Server-Side Attacks
| File | Vulnerability |
|------|--------------|
| `ssrf.md` | SeExcerpt of 6,738 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:421b4f2958320039, llm:description: 'A SAST skill that gives AI coding agents structured vulnerability detection across 34 vulnerability classes.'
matched fp:421b4f2958320039, llm:description: 'A SAST skill that gives AI coding agents structured vulnerability detection across 34 vulnerability classes.'
matched fp:421b4f2958320039, llm:description: 'A SAST skill that gives AI coding agents structured vulnerability detection across 34 vulnerability classes.'