Audit candidate: Manavarya09/design-extract

[xiaolai]

Repo

https://github.com/Manavarya09/design-extract

Field	Value
Stars	2482
Source	manual queue (2026-05-12 discovery batch)
Description	Extract any website's complete design system with one command. DTCG tokens, semantic+primitive+composite, MCP server for Claude Code/Cursor/Windsurf, multi-platform emitters (iOS SwiftUI, Android Comp

Next Steps

1. Review the repo — check CONTRIBUTING.md, recent activity, PR history
2. If worthy: add label audit-ready to trigger the audit workflow
3. If not: close this issue

Checklist

• Accepts contributions
• Updated in last 60 days
• Not a fork or mirror
• Has meaningful NL artifacts

[github-actions]

Audit Complete: Manavarya09/design-extract

NL Score: 94/100 | Security: BLOCKED | Artifacts: 11 | Strategy: single

NLPM Audit: Manavarya09/design-extract

Date: 2026-04-06 | Artifacts: 11 | Strategy: single
NL Score: 94/100
Security: BLOCKED
Bugs: 0 | Quality Issues: 15 | Security Findings: 16

NL Score Summary

File	Type	Score	Top Issue
commands/extract.md	command	83	Multi-step post-run flow uses bullets, not numbered steps; vague "tight summary"
commands/remix.md	command	85	Multi-step post-run flow in prose, no numbered steps
commands/pair.md	command	93	Missing allowed-tools; vague "most distinctive crossover"
commands/battle.md	command	95	Missing allowed-tools
commands/brand.md	command	95	Missing allowed-tools
commands/grade.md	command	95	Missing allowed-tools
commands/pack.md	command	95	Missing allowed-tools
commands/theme-swap.md	command	95	Missing allowed-tools
skills/extract-design/SKILL.md	skill	96	Vague quantifiers: "key findings", "notable design decisions"
website/CLAUDE.md	context	98	Vague "relevant guide" in included AGENTS.md
.claude-plugin/plugin.json	manifest	100	—

Security Scan

Severity	Count
Critical	0
High	9
Medium	1
Low	6

Execution Surface Inventory

Surface	Files
Hooks	0
Scripts (.sh)	1 — raycast-extension/scripts/dev-setup.sh
Scripts (.js, .py)	111+ in src/, 1 in bin/ (not individually scanned)
MCP configs	0
Package manifests	1 — package.json

Security Findings

#	Severity	File	Line	Pattern	Description
1	HIGH	package.json	11	SEC-postinstall-script	postinstall auto-runs npx playwright install chromium --with-deps on every npm install; --with-deps triggers OS-level system package installation (apt/brew) and may invoke sudo on Linux
2	HIGH	commands/battle.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang battle $ARGUMENTS without quoting or sanitization; malformed or adversarial input can inject arbitrary shell commands
3	HIGH	commands/brand.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang brand $ARGUMENTS without quoting or sanitization
4	HIGH	commands/extract.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang $ARGUMENTS without quoting or sanitization
5	HIGH	commands/grade.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang grade $ARGUMENTS without quoting or sanitization
6	HIGH	commands/pack.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang pack $ARGUMENTS without quoting or sanitization
7	HIGH	commands/pair.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang pair $ARGUMENTS without quoting or sanitization
8	HIGH	commands/remix.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang remix $ARGUMENTS without quoting or sanitization
9	HIGH	commands/theme-swap.md	9	SEC-shell-injection	$ARGUMENTS interpolated into npx designlang theme-swap $ARGUMENTS without quoting or sanitization

---

Full report: https://github.com/xiaolai/nlpm-for-claude/blob/main/auditor/audits/Manavarya09-design-extract.md

SECURITY BLOCKED: Critical/High security issues found. Do NOT submit automated PRs. Manual security review required.

[github-actions]

Security disclosure auto-filing failed (default GITHUB_TOKEN cannot create issues on third-party repos). Body queued at auditor/disclosures-pending/Manavarya09-design-extract.md for manual filing.