ClawHub
Back to skill

Security audit

ad-creative

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a legitimate ad-creative helper, but it includes under-scoped instructions that could install/run third-party code and affect ad operations if followed too freely.

Review before installing. Treat it as an ad-creative drafting skill, not an autonomous campaign operator. Do not let an agent run the npx/gooseworks/create-video, Remotion, or git clone/make commands unless versions or commits are pinned and the work runs in a sandbox. Require explicit approval before campaign creation or upload, use only authorized customer/prospect data and consented voices, and fix the review-template escaping guidance before hosting pages for clients.

A.I.G

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T08 · Insecure Dependencies

Error
Location
references/imessage-video-ads.md:80
Finding

Unpinned Third-Party Installation Commands Allow Supply-Chain Code Execution

Content
View full analysis
Remediation
View remediation
install --all npx --yes create-video@ ``` 2. Pin source installations to an immutable reviewed commit: ```bash git clone https://github.com/jamiepine/voicebox.git cd voicebox git checkout --detach ``` 3. Publish the expected package integrity digest or commit hash in the reference documentation and require verification before execution. 4. Require inspection of package lifecycle scripts, Makefiles, and dependency manifests before running installation commands. 5. Use lockfiles with integrity metadata and immutable or frozen installation modes where supported. 6. Run third-party installation and rendering tools in an isolated, least-privileged environment such as a disposable container or virtual machine. Do not expose unrelated host directories, SSH agents, cloud metadata endpoints, or unnecessary credentials. 7. Remove `@latest` from all executable examples. Version updates should undergo a new review before documentation is changed. 8. Clearly distinguish informational examples from commands an agent is authorized to execute. Require explicit user approval before downloading or executing third-party code. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
assets/creative-review-template.html:2
Finding

Incorrect JSON Escaping Guidance Can Enable Stored Script Injection in Generated Review Pages

Content
View full analysis
below with the real project. Everything else renders from it. The file is self-contained — no build, no network, no dependencies. Open it in a browser, host it on any static host (Vercel/Netlify/GitHub Pages), or hand off the single .html file. THE DATA BLOCK IS JSON, NOT JAVASCRIPT: - double-quoted keys and strings, no comments, no trailing commas - it is inert data (parsed with JSON.parse), so a value can never execute - SECURITY: escape every literal "<" in your text values as < so a value like "" can never break out of the tag. All values are also HTML-escaped again at render time. --> ``` The data is embedded directly in a raw-text script element at `assets/creative-review-template.html:147-148`: ```html
Remediation
View remediation
``` The test should verify that no additional script element is created and that the text renders only as inert content. 4. Retain the existing render-time `esc()` function. Source-level JSON escaping and render-time HTML escaping protect different parser boundaries and are both required. 5. Add a restrictive Content Security Policy. Prefer moving the rendering JavaScript into a separate static file so inline scripts can be disabled. If the single-file requirement must remain, use a fixed script hash rather than allowing unrestricted inline execution. 6. Validate the completed artifact before delivery by parsing it in a browser automation test and checking the number and type of script elements. 7. Keep review pages on a dedicated static origin without sensitive cookies or privileged same-origin APIs, limiting the impact of any future client-side injection defect. ]]>

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (23)

Harmful Content Injection

Critical
Category
Prompt Injection
Confidence
70% confidence
Finding

This content may contain harmful instructions that could cause physical harm if followed. CRITICAL: Review carefully before use.

Content

Scanner excerpt · assets/creative-review-template.html (reported line 169)May include surrounding context.

html
],
      "frames": [
        { "label": "Hook", "prompt": "Product bag hero on soft pink, gold-lace overlay", "headline": "Finally — a plant-based protein that's third-party tested for heavy metals.", "headlineTheme": "dark" },
        { "label": "The problem", "prompt": "Editorial card: 'Plants absorb more than nutrients' + Pb/As/Cd chips" },
        { "label": "Enter Light Labs", "prompt": "Clean card: 'So we sent it to Light Labs' + independent-lab note" },
        { "label": "The results", "prompt": "Results table: Arsenic / Cadmium / Lead, all within limits, green check" },
        { "label": "For context", "prompt": "'Less arsenic than your breakfast' comparison bar" },
        { "label": "The ask", "prompt": "Product you can finally trust — CTA frame", "headline": "Protein you can finally trust." }
      ],
      "headlines": [
        "Finally — a plant-based protein that's third-party tested for heavy metals.",

Harmful Content Injection

Critical
Category
Prompt Injection
Confidence
70% confidence
Finding

This content may contain harmful instructions that could cause physical harm if followed. CRITICAL: Review carefully before use.

Content

Scanner excerpt · assets/creative-review-template.html (reported line 170)May include surrounding context.

html
"frames": [
        { "label": "Hook", "prompt": "Product bag hero on soft pink, gold-lace overlay", "headline": "Finally — a plant-based protein that's third-party tested for heavy metals.", "headlineTheme": "dark" },
        { "label": "The problem", "prompt": "Editorial card: 'Plants absorb more than nutrients' + Pb/As/Cd chips" },
        { "label": "Enter Light Labs", "prompt": "Clean card: 'So we sent it to Light Labs' + independent-lab note" },
        { "label": "The results", "prompt": "Results table: Arsenic / Cadmium / Lead, all within limits, green check" },
        { "label": "For context", "prompt": "'Less arsenic than your breakfast' comparison bar" },
        { "label": "The ask", "prompt": "Product you can finally trust — CTA frame", "headline": "Protein you can finally trust." }
      ],
      "headlines": [
        "Finally — a plant-based protein that's third-party tested for heavy metals.",
        "We tested our protein for heavy metals. Here's what an independent lab found.",

Harmful Content Injection

Critical
Category
Prompt Injection
Confidence
70% confidence
Finding

This content may contain harmful instructions that could cause physical harm if followed. CRITICAL: Review carefully before use.

Content

Scanner excerpt · assets/creative-review-template.html (reported line 196)May include surrounding context.

html
"frames": [
        { "label": "Hook", "prompt": "Product bag hero on soft pink, gold-lace overlay", "headline": "Finally — a plant-based protein that's third-party tested for heavy metals.", "headlineTheme": "dark" },
        { "label": "The problem", "prompt": "Editorial card: 'Plants absorb more than nutrients' + Pb/As/Cd chips" },
        { "label": "Enter Light Labs", "prompt": "Clean card: 'So we sent it to Light Labs' + independent-lab note" },
        { "label": "The results", "prompt": "Results table: Arsenic / Cadmium / Lead, all within limits, green check" },
        { "label": "For context", "prompt": "'Less arsenic than your breakfast' comparison bar" },
        { "label": "The ask", "prompt": "Product you can finally trust — CTA frame", "headline": "Protein you can finally trust." }
      ],
      "headlines": [
        "Finally — a plant-based protein that's third-party tested for heavy metals.",
        "We tested our protein for heavy metals. Here's what an independent lab found.",

Harmful Content Injection

Critical
Category
Prompt Injection
Confidence
70% confidence
Finding

This content may contain harmful instructions that could cause physical harm if followed. CRITICAL: Review carefully before use.

Content

Scanner excerpt · assets/creative-review-template.html (reported line 203)May include surrounding context.

html
"frames": [
        { "label": "Hook", "prompt": "Product bag hero on soft pink, gold-lace overlay", "headline": "Finally — a plant-based protein that's third-party tested for heavy metals.", "headlineTheme": "dark" },
        { "label": "The problem", "prompt": "Editorial card: 'Plants absorb more than nutrients' + Pb/As/Cd chips" },
        { "label": "Enter Light Labs", "prompt": "Clean card: 'So we sent it to Light Labs' + independent-lab note" },
        { "label": "The results", "prompt": "Results table: Arsenic / Cadmium / Lead, all within limits, green check" },
        { "label": "For context", "prompt": "'Less arsenic than your breakfast' comparison bar" },
        { "label": "The ask", "prompt": "Product you can finally trust — CTA frame", "headline": "Protein you can finally trust." }
      ],
      "headlines": [
        "Finally — a plant-based protein that's third-party tested for heavy metals.",
        "We tested our protein for heavy metals. Here's what an independent lab found.",

Harmful Content Injection

Critical
Category
Prompt Injection
Confidence
70% confidence
Finding

This content may contain harmful instructions that could cause physical harm if followed. CRITICAL: Review carefully before use.

Content

Scanner excerpt · assets/creative-review-template.html (reported line 206)May include surrounding context.

html
"frames": [
        { "label": "Hook", "prompt": "Product bag hero on soft pink, gold-lace overlay", "headline": "Finally — a plant-based protein that's third-party tested for heavy metals.", "headlineTheme": "dark" },
        { "label": "The problem", "prompt": "Editorial card: 'Plants absorb more than nutrients' + Pb/As/Cd chips" },
        { "label": "Enter Light Labs", "prompt": "Clean card: 'So we sent it to Light Labs' + independent-lab note" },
        { "label": "The results", "prompt": "Results table: Arsenic / Cadmium / Lead, all within limits, green check" },
        { "label": "For context", "prompt": "'Less arsenic than your breakfast' comparison bar" },
        { "label": "The ask", "prompt": "Product you can finally trust — CTA frame", "headline": "Protein you can finally trust." }
      ],
      "headlines": [
        "Finally — a plant-based protein that's third-party tested for heavy metals.",
        "We tested our protein for heavy metals. Here's what an independent lab found.",

Harmful Content Injection

Critical
Category
Prompt Injection
Confidence
70% confidence
Finding

This content may contain harmful instructions that could cause physical harm if followed. CRITICAL: Review carefully before use.

Content

Scanner excerpt · assets/creative-review-template.html (reported line 208)May include surrounding context.

html
"frames": [
        { "label": "Hook", "prompt": "Product bag hero on soft pink, gold-lace overlay", "headline": "Finally — a plant-based protein that's third-party tested for heavy metals.", "headlineTheme": "dark" },
        { "label": "The problem", "prompt": "Editorial card: 'Plants absorb more than nutrients' + Pb/As/Cd chips" },
        { "label": "Enter Light Labs", "prompt": "Clean card: 'So we sent it to Light Labs' + independent-lab note" },
        { "label": "The results", "prompt": "Results table: Arsenic / Cadmium / Lead, all within limits, green check" },
        { "label": "For context", "prompt": "'Less arsenic than your breakfast' comparison bar" },
        { "label": "The ask", "prompt": "Product you can finally trust — CTA frame", "headline": "Protein you can finally trust." }
      ],
      "headlines": [
        "Finally — a plant-based protein that's third-party tested for heavy metals.",
        "We tested our protein for heavy metals. Here's what an independent lab found.",

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 329)May include surrounding context.

md
cision made off one link instead of by reading markdown. The template ships at [assets/creative-review-template.html](assets/creative-review-template.html) (one

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
80% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · SKILL.md (reported line 376)May include surrounding context.

md
## Common Mistakes

- **Writing headlines that only work together** — RSA headlines get combined randomly
- **Ignoring character limits** — Platforms truncate without warning
- **All variations sound the same** — Vary angles, not just word choice
- **No CTA headlines** — RSAs need action-oriented headlines to drive clicks; include at least 2-3
- **Generic descriptions** — "Learn more about our solution" wastes the slot

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · assets/creative-review-template.html (reported line 2)May include surrounding context.

html
<!DOCTYPE html>
<!--
  Creative Review Page — a shareable ad-creative approval artifact.

  HOW TO USE (agents): replace the JSON inside <script id="review-data"> below

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation description contains many broad trigger phrases such as general requests for ads, strategy, hook writing, review pages, and multiple creative formats. Overbroad routing can cause this skill to activate for requests outside its intended boundaries, pulling in irrelevant instructions or bypassing more appropriate, narrower skills, which is especially risky because this file also references adjacent operational capabilities.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill is scoped as ad-creative generation, but the Tool Integrations section includes campaign management and creation actions, which expands it into operational ad execution. This creates a privilege/scope mismatch: a user invoking copy generation could be steered into modifying live campaigns or creating new ones, increasing the chance of unintended spend, account changes, or abuse if the agent has tool access.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This JSON eval file is a manifest-type file, so vague-trigger checks apply. The expectation that the skill 'Should trigger on casual phrasing' and the assertion 'Triggers on casual phrasing' do not define boundaries, exclusions, or specific trigger phrases, making activation criteria overly broad and prone to unintended invocation from ordinary ad-related requests.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The document explicitly recommends "scraping" and competitor profiling as inputs to ad strategy without any accompanying guardrails about consent, terms-of-service, robots.txt, rate limits, copyright, privacy, or platform-policy compliance. In an agent skill, this can normalize or trigger unauthorized collection of third-party content and personal data, creating legal, privacy, and account-enforcement risk even if the underlying business goal is legitimate.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The document recommends brand voice cloning tools without warning that cloning a person's voice generally requires explicit consent and may trigger legal, privacy, publicity-right, fraud, and platform-policy issues. In an ad-creative skill, that omission is more dangerous because it normalizes production use of deceptive or unauthorized synthetic voices at scale.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The reference promotes personalized video generation using names and data but omits privacy, consent, retention, and data-minimization warnings. In advertising workflows, this can encourage users to process prospect or customer data in third-party AI tools without appropriate legal basis or contractual safeguards.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/generative-tools.md (reported line 50)May include surrounding context.

API example:

bash
# Using the Gemini API for image generation
curl -X POST "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash-image:generateContent" \
  -H "Content-Type: application/json" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/generative-tools.md (reported line 322)May include surrounding context.

API example:

bash
curl -X POST "https://api.elevenlabs.io/v1/text-to-speech/{voice_id}" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/generative-tools.md (reported line 322)May include surrounding context.

API example:

bash
curl -X POST "https://api.elevenlabs.io/v1/text-to-speech/{voice_id}" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The section on personalized outreach videos specifically encourages addressing prospects by name without any privacy or consent caveat. Because this is directly tied to outbound advertising/sales use, it raises tangible risk of unlawful profiling, unauthorized data sharing, or intrusive targeting practices.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The document instructs users to run npx gooseworks install --all without pinning a package version or verifying package integrity. Because npx resolves and executes code from the registry at run time, a compromised package, typosquat, or malicious new release could result in arbitrary code execution on the user's machine or agent environment.

Content

No source excerpt is available for this finding.