RipLock

Security scanner — we lock it down so you can let it rip.

RipLock scans your codebase for security vulnerabilities using AST taint tracking, structural pattern matching, and 537 security checks across 7 languages and 10+ infrastructure platforms. It also scans packages before you install them to catch supply chain attacks. Every finding comes with a plain-English explanation and copy-paste fix.

Quick Start

npx riplock-cli-cli                            # Scan your code
npx riplock-cli-cli scan-pkg express axios     # Scan packages before installing

How It Works

RipLock uses five detection engines:

1. AST Taint Tracking — Parses code with tree-sitter, traces user input through variables, object properties, promise callbacks, and function calls to dangerous sinks. Cross-file analysis follows imports 3 levels deep.
2. ast-grep Structural Matching — Semgrep-style pattern-as-code matching with metavariables ($DB.query($SQL)) and compound operators (inside, has, not).
3. 268 Hardcoded Checks — Battle-tested TypeScript checks with custom validation logic, iteratively calibrated against 18 real-world projects.
4. 269 Rule Engine Rules — JSON-defined patterns (regex and ast-grep) loaded at runtime. User-extensible via .riplock-rules.json.
5. Cross-File Taint — Resolves imports between project files, builds function taint signatures, detects data flow across module boundaries.

// RipLock traces this across variables and files:
// file: routes/users.ts
import { findUser } from './db';
app.get('/users', async (req, res) => {
  const id = req.params.id;           // ← source
  const user = await findUser(id);     // ← cross-file call
  res.json(user);
});

// file: db.ts
export function findUser(id) {
  return db.query(`SELECT * FROM users WHERE id = ${id}`);  // ← sink
}

Output:

CRITICAL  SQL Injection (Taint-Tracked)                AST-INJ001
  Data flow:
  └ req.params.id (line 3)
  ├ id (line 4)
  └ db.query(...) (line 9, db.ts)

Supported Languages

Supported Infrastructure

Terraform, Kubernetes, Helm, CloudFormation, Docker, Nginx, Apache, Ansible, Serverless Framework, GitHub Actions — plus 12 generic config checks for any YAML/JSON/TOML/INI file.

Usage

Scan your code

npx riplock-cli                           # Scan current directory
npx riplock-cli ./my-project              # Scan a specific path
npx riplock-cli --severity high           # Only critical + high findings
npx riplock-cli --json                    # JSON output for CI
npx riplock-cli --sarif                   # SARIF 2.1.0 for GitHub Code Scanning
npx riplock-cli --ignore SEC007 GIT010    # Skip specific checks
npx riplock-cli --exclude 'tests/**'      # Exclude files by glob pattern
npx riplock-cli --no-deps                 # Skip npm audit (faster)
npx riplock-cli --scan-deps               # Scan installed dependencies (node_modules etc.)
npx riplock-cli --list-checks             # Show all checks
npx riplock-cli --verbose                 # Show timing and debug info

Scan packages before installing

npx riplock-cli scan-pkg express@4.21.2           # Scan an npm package
npx riplock-cli scan-pkg requests==2.31.0 --pip   # Scan a pip package
npx riplock-cli scan-pkg lodash moment axios      # Scan multiple packages
npx riplock-cli scan-pkg express --json           # JSON output for automation

Downloads the package tarball without installing, extracts it to a temp directory, runs 12 supply chain checks (credential exfiltration, reverse shells, install script attacks, obfuscated code, cryptominers, etc.), and cleans up. Use this to vet packages before adding them to your project.

What It Catches

537 checks across 18 categories:

| Supply Chain | 12 | Credential exfiltration, reverse shells, install script attacks, obfuscated code, cryptominers, .pth code execution, DNS exfiltration, system persistence |

Run npx riplock-cli --list-checks for the full catalog.

Custom Rules

Create .riplock-rules.json in your project:

{
  "rules": [
    {
      "id": "CUSTOM-001",
      "message": "Direct database access outside repository layer",
      "severity": "high",
      "category": "injection",
      "languages": ["typescript"],
      "pattern": "prisma.$TABLE.$METHOD",
      "pattern-mode": "ast",
      "paths": { "exclude": ["src/repositories/**"] },
      "fix": "Use the repository pattern for all database access"
    }
  ]
}

Supports regex patterns, ast-grep structural patterns with $METAVAR metavariables, and compound operators (inside, has, not, follows, precedes).

Configuration

Create .riplock.json in your project root:

{
  "disable": ["CONFIG011", "CICD002"],
  "severity": "medium",
  "exclude": ["tests/**", "scripts/**"],
  "skipDeps": true
}

CI Integration

GitHub Actions

- name: Security scan
  run: npx riplock-cli --severity high

# With SARIF upload to GitHub Code Scanning:
- name: Security scan (SARIF)
  run: npx riplock-cli --sarif > results.sarif
  continue-on-error: true
- name: Upload SARIF
  uses: github/codeql-action/upload-sarif@v3
  if: always()
  with:
    sarif_file: results.sarif

GitLab CI

security_scan:
  script: npx riplock-cli --severity high --json > riplock.json
  artifacts:
    reports:
      security: riplock.json

Using with AI Coding Assistants

RipLock scans every file systematically in milliseconds — something that would cost an AI assistant 200K+ tokens to do by reading files one at a time. Run RipLock first, then let your AI focus on the findings that need judgment.

Option 1: Manual

Run it yourself and share findings with your AI:

riplock . --severity high
# Copy interesting findings into your AI conversation

# Or pipe JSON directly:
riplock . --json | pbcopy
# Then: "Here are the security findings — fix them"

Option 2: AI instructions (recommended)

Add to your project's CLAUDE.md, .cursorrules, or equivalent instructions file:

## Security

Before committing or creating a PR, run a security scan:
  riplock . --no-deps --severity high

Fix any critical or high findings before proceeding.

Your AI assistant will run the scan as part of its workflow, read the findings, and fix them with full context of what the code is supposed to do. This is the best balance of coverage and simplicity.

Option 3: Pre-install hook (Claude Code)

RipLock ships ready-made Claude Code hooks in the hooks/ directory. The pre-install hook scans packages with riplock scan-pkg before allowing npm install, pip install, npx -y, or uvx commands. It also hard-blocks ephemeral package runners (npx -y, uvx) without version pins, since those have no lockfile.

Copy the hooks and add to your .claude/settings.json:

{
  "hooks": {
    "PreToolUse": [{
      "matcher": "Bash",
      "hooks": [{
        "type": "command",
        "command": "/path/to/hooks/check-install-safety.sh",
        "timeout": 60,
        "statusMessage": "Scanning packages for malware..."
      }]
    }],
    "PostToolUse": [{
      "matcher": "Write|Edit",
      "hooks": [{
        "type": "command",
        "command": "/path/to/hooks/check-supply-chain.sh",
        "timeout": 10,
        "statusMessage": "Checking for supply chain risks..."
      }]
    }]
  }
}

The pre-install hook downloads and scans packages before installation. The post-write hook catches supply chain risks in config files (unpinned MCP servers, */latest dependencies, install scripts).

Option 4: Post-edit scan hook

Surface findings as warnings after every code change:

{
  "hooks": {
    "PostToolUse": [{
      "matcher": "Write|Edit",
      "hooks": [{
        "type": "command",
        "command": "riplock . --no-deps --severity critical --json 2>/dev/null | node -e \"const d=JSON.parse(require('fs').readFileSync('/dev/stdin','utf8')); if(d.stats.critical>0) console.log('⚠️ RipLock: '+d.stats.critical+' critical findings')\""
      }]
    }]
  }
}

Option 5: Claude Code skill (most powerful)

Create ~/.claude/skills/security-review.md to get a /security-review slash command that combines RipLock's systematic scan with Claude's contextual review:

---
name: security-review
description: Combined static + contextual security review
user-invocable: true
---

# Security Review

## Phase 1: RipLock Static Analysis

Run RipLock against the current project:

\`\`\`bash
npx riplock-cli . --no-deps --json
\`\`\`

Parse the JSON output. Summarize findings by severity with the top 5 most
important issues, their taint flow paths, and fix suggestions.

## Phase 2: Contextual Review

After RipLock completes, review the codebase for issues static analysis misses:

1. **Business Logic** — payment flows, authorization bypasses, workflow ordering
2. **Auth Architecture** — middleware ordering, inconsistent auth, privilege escalation
3. **Data Exposure** — over-fetched API responses, server data leaking to client
4. **Architectural Risks** — secrets management, missing rate limiting, error handling

Read the key files and report contextual issues with specific fix suggestions.
End with a prioritized action list.

Then run /security-review in any project. RipLock handles the O(n) file scanning, Claude handles the contextual judgment.

Exit Codes

Scoring

Findings with 4+ occurrences are automatically grouped with the top 3 examples shown. Each grouped finding includes a suppress hint.

Limitations

RipLock is a first line of defense, not a guarantee. It catches known vulnerability patterns and common supply chain attack techniques, but:

• It is not a substitute for professional security review. Static analysis finds what it's been taught to look for. Business logic flaws, novel attack techniques, and subtle backdoors require human judgment.
• A clean scan-pkg result does not mean a package is safe. It means the package didn't match any of the 12 supply chain patterns RipLock checks. Sophisticated attackers can craft payloads that evade pattern matching.
• Cross-file taint tracking has depth limits — call chains beyond 3 import levels are not followed.
• Typosquatting is not detected — telling lod-ash from lodash requires registry metadata, not code analysis.
• Runtime behavior is not analyzed — memory leaks, race conditions, and timing attacks require dynamic testing.
• Calibration is ongoing. The check thresholds have been tuned against ~20 real-world projects but have not yet been battle-tested across a wide range of production codebases. If you hit false positives, open an issue or suppress individual checks in .riplock.json.

Use RipLock alongside — not instead of — dependency auditing (npm audit, Dependabot/Renovate), lockfile enforcement, code review, and periodic professional penetration testing.

License

MIT