Back to skill

Security audit

gold-radar

Security checks for vulnerabilities and agentic risk

Overview

This is not malware, but it needs review because it gives broad, prescriptive gold-trading guidance and carries an unnecessary npm dependency chain.

Install only if you specifically want gold-market analysis and are comfortable with finance-oriented recommendations. Treat all trading outputs as informational, not personalized financial advice, and review the npm dependency metadata before installing, especially the third-party mirror and unnecessary dependency chain.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T08 · Insecure Dependencies

Note
Location
package.json:2
Finding

Unnecessary Runtime Dependency Retrieved Through a Third-Party Package Mirror

Content
View full analysis

Vulnerability Details

File Location: package.json:2-4; package-lock.json:8,89-91
Vulnerability Type: Unnecessary third-party dependency and unsafe package source
Risk Level: Low

Vulnerable Code

package.json:2-4:

json
"dependencies": {
  "clawhub": "^0.9.0"
}

package-lock.json:89-91:

json
"node_modules/clawhub": {
  "version": "0.9.0",
  "resolved": "https://registry.npmmirror.com/clawhub/-/clawhub-0.9.0.tgz",

Technical Analysis

The project contains documentation and reference material only. No executable project code uses the clawhub package, and the declared gold-market analysis workflows do not require this runtime dependency. Installing it therefore exceeds the minimum dependency footprint necessary for the Skill's functionality.

The lockfile resolves clawhub and its transitive dependencies through registry.npmmirror.com, a third-party mirror rather than the official npm registry. Package integrity hashes provide protection against content differing from the committed lockfile, but they do not remove the need to trust the lockfile itself or the original dependency selection. The semver declaration ^0.9.0 can also permit later compatible releases when dependency resolution is performed without strictly honoring the existing lockfile.

No lifecycle install hook or malicious package code was identified in the audited project files. Consequently, this is a supply-chain exposure rather than evidence that malicious execution currently occurs.

Attack Path

  1. A user or deployment process runs npm install or an equivalent dependency installation command.
  2. npm reads the unnecessary clawhub dependency and its transitive dependency graph.
  3. Package archives are retrieved from the third-party mirror specified by the lockfile.
  4. If the dependency, a transitive package, the mirror response, or a modified lockfile is compromised, attacker-controlled pa ...[truncated 911 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove clawhub from package.json because no project code or documented workflow requires it.
  2. Delete and regenerate package-lock.json after removing the dependency, or remove package metadata entirely if the Skill is documentation-only.
  3. If the package is genuinely required later, document the exact functionality that depends on it and pin a reviewed exact version rather than using a caret range.
  4. Resolve packages from an organization-approved registry, preferably the official npm registry or a controlled internal proxy.
  5. Use deterministic installation with npm ci and retain integrity-protected lockfiles under review.
  6. Disable lifecycle scripts during installation where compatible, such as with npm ci --ignore-scripts.
  7. Add dependency review, vulnerability scanning, and lockfile-diff checks to the release process.
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (11)

Known Vulnerable Dependency: undici==7.24.5 — 12 advisory(ies): CVE-2026-6733 (undici vulnerable to HTTP response queue poisoning via keep-alive socket reuse); CVE-2026-13697 (undici vulnerable to cross-user information disclosure and parse-time crash via ); CVE-2026-16728 (undici vulnerable to downstream response desynchronization via retry interceptor) +9 more

High
Category
Supply Chain
Confidence
97% confidence
Finding

The lockfile includes undici 7.24.5, which the finding reports as affected by multiple high-severity HTTP client issues including response queue poisoning, information disclosure, and response desynchronization. Because this skill is explicitly a real-time market monitoring and alerting tool, network communication is core to its functionality, so an HTTP client flaw is especially relevant: a malicious or compromised upstream service, proxy, or intermediary could corrupt request/response handling, leak data across sessions, or cause incorrect market data to be consumed.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The activation description is broad enough to trigger on general investing, market outlook, or portfolio-management discussions that may not specifically require this skill. In a finance context, overbroad routing is risky because it can cause the agent to deliver trading-oriented guidance or alerts in conversations where the user did not explicitly request actionable financial decision support.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The invocation examples are ambiguous and encourage activation for loosely related topics like economic indicators, trading advice, or portfolio tracking without defining exclusion conditions. This increases the chance of accidental invocation and inappropriate financial guidance, especially when the broader assistant may have safer or more general-purpose handling for informational market questions.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill provides trading signals, alert levels, and recommended actions including language like 'execute trading decisions' and 'should I buy/sell,' which can directly influence user asset decisions. Although a disclaimer exists later, it is weak relative to the operational guidance, and in a high-stakes financial setting this can lead users to act on unverified or unsuitable recommendations with real monetary loss.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill description defines very broad activation criteria across general gold prices, investment analysis, trading signals, macro events, and portfolio management. That increases the chance of unintended invocation in ordinary finance conversations, causing the agent to enter a high-impact advisory workflow and potentially provide trading-oriented guidance when the user did not explicitly request this skill.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The invocation examples and signal phrases around buy/sell checks are ambiguous because they map short, common questions like 'should I buy?' or 'sell now?' to this skill without requiring clear gold-specific context. This can misfire on unrelated asset discussions and lead the system to generate actionable investment recommendations in the wrong context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The file contains concrete investment guidance, including entry/exit conditions, position sizing, stop-loss thresholds, and monthly allocation recommendations, but does not include a clear user-facing disclaimer that the content is not financial advice and may not be suitable for all users. In a skill explicitly designed for real-time gold monitoring and investment decision support, users are likely to treat this as actionable advice, increasing the chance of financial harm from overreliance on the guidance.

Content

No source excerpt is available for this finding.

Known Vulnerable Dependency: fflate==0.8.2 — 1 advisory(ies): CVE-2026-45820 (fflate unzipSync can enter an infinite loop when parsing malformed ZIP64 archive)

Low
Category
Supply Chain
Confidence
89% confidence
Finding

The lockfile pins fflate to 0.8.2, which is reported as vulnerable to an infinite loop when processing a malformed ZIP64 archive. If this skill or its dependency chain ever decompresses attacker-controlled archives, an attacker could trigger a denial of service by causing the process to hang or consume CPU. In this gold-monitoring context the issue is somewhat less dangerous than in a file-ingestion tool, but the presence of a package used by a generic CLI/deployment utility still makes the risk real.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
88% confidence
Finding

The dependency uses a caret range (^0.9.0), which allows automatic installation of newer compatible versions within the 0.x range. This increases supply-chain risk because builds are not deterministic and a newly published package version could introduce malicious code or breaking behavior without any change to this skill's source.

Content

Scanner excerpt · package.json (reported line 3)May include surrounding context.

json
{
  "dependencies": {
    "clawhub": "^0.9.0"
  }
}

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

This markdown file consists solely of Chinese headings and descriptions, which can amount to a language/locale constraint in the skill content. The policy allows language restrictions only when the skill offers user choice or clearly documents and justifies the locale limitation, which is not present here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The entire markdown document is written in Chinese and provides no indication that the skill supports language selection or that it is intentionally restricted to a Chinese-speaking or China-specific audience. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.