Back to skill

Security audit

Global Video Title Generator

Security checks for vulnerabilities and agentic risk

Overview

This is mainly a local video-title/content generation package with messy prototype and marketing materials, but I found no artifact-backed exfiltration, persistence, destructive behavior, or hidden privilege use.

Install only if you are comfortable with a prototype-quality local content generator. Do not rely on its advertised quotas, payments, multilingual support, trend analysis, or compliance claims without separate verification, and avoid running setup.py or publish.py unless you intend to let it create or rewrite local project metadata files.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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 (3)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
scripts/main.py:44
Finding

Paid-Tier Authorization Can Be Forged Using an API Key Prefix

Content
View full analysis
str: """Determine user tier based on API key""" if not self.api_key: return "free" # In a real implementation, this would check against a database # For now, we'll simulate based on API key format if self.api_key.startswith("basic_"): return "basic" elif self.api_key.startswith("pro_"): return "pro" elif self.api_key.startswith("enterprise_"): return "enterprise" else: return "free" ``` The complete-content generator implements the same vulnerable authorization mechanism: ```python def _determine_user_tier(self) -> str: """Determine user tier""" if not self.api_key: return "free" if self.api_key.startswith("basic_"): return "basic" elif self.api_key.startswith("pro_"): return "pro" elif self.api_key.startswith("enterprise_"): return "enterprise" else: return "free" ``` ### Technical Analysis The application treats an API key's attacker-controlled prefix as proof of subscription status. It does not verify the key against a trusted datastore, validate a cryptographic signature, check expiration, or confirm that the key belongs to an authorized account. Consequently, the API key is functioning as an untrusted role selector rather than an authentication credential. Any caller who knows the documented naming convention can select a paid tier by supplying an arbitrary string beginning with `basic_`, `pro_`, or `enterprise_`. The enterprise tier is especially significant because downstream quota checks explicitly grant it unrestricted access: ```python if self.user_tier == "enterprise": return True ``` ### Attack Path 1. An attacker ...[truncated 1305 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/main.py:75
Finding

Usage Quotas Can Be Bypassed Through Unbounded or Negative Generation Counts

Content
View full analysis
bool: """Check if user has reached usage limit""" if self.user_tier == "enterprise": return True # No limits for enterprise daily_usage = self.usage_tracker.get_daily_usage(user_id) if self.user_tier == "free": limit = self.config.get("free_tier", {}).get("daily_generations", 5) elif self.user_tier == "basic": limit = 500 // 30 elif self.user_tier == "pro": limit = 2000 // 30 else: limit = 5 return daily_usage < limit ``` The CLI accepts any integer without requiring it to be positive or imposing an upper bound: ```python parser.add_argument("--count", type=int, default=5, help="Number of titles to generate") ``` The complete-content generator calculates credits directly from t ...[truncated 3397 chars]
Remediation
View remediation
max_batch: return error requested_cost = cost_per_item * requested_count if current_usage + requested_cost > quota: return error ``` 4. Prevent usage counters from becoming negative and reject any attempted decrement outside a dedicated refund workflow. 5. Make quota checking and quota reservation atomic so concurrent requests cannot both pass using the same remaining balance. 6. Persist quota data in a trusted datastore rather than generator-instance memory. 7. Include the date or billing period in credit-storage keys and reset quotas through controlled period transitions. 8. Reserve credits before generation and either finalize or safely refund the reservation after completion. 9. Add automated tests for: - Negative counts. - Zero counts. - Counts equal to and greater than the remaining quota. - Extremely large counts. - Concurrent requests. - Counter underflow. 10. Apply the same centralized validation and quota service to both generator implementations. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/main.py:269
Finding

Process-Local Usage Tracking Allows Quota Reset by Recreating the Generator

Content
View full analysis
date -> count self.total_usage = {} # user_id -> total count def record_usage(self, user_id: str, count: int, tier: str): """Record usage for a user""" if not user_id: user_id = "anonymous" today = date.today().isoformat() key = f"{user_id}_{today}" if key not in self.daily_usage: self.daily_usage[key] = 0 self.daily_usage[key] += count if user_id not in self.total_usage: self.total_usage[user_id] = 0 self.total_usage[user_id] += count def get_daily_usage(self, user_id: str) -> int: """Get today's usage for a user""" if not user_id: user_id = "anonymous" today = date.today().isoformat() key = f"{user_id}_{today}" return self.daily_usage.get(key, 0) ``` The complete generator's credit counter is also created lazily on the individual object: ```python if not hasattr(self, '_credit_usage'): self._credit_usage = {} key = f"{user_id}_credits" return self._credit_usage.get(key, 0) ``` ### Technical Analysis Usa ...[truncated 1594 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Rogue AgentSelf-Modification, Session Persistence
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (69)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

Business-model validation, revenue calculations, roadmap/config inspection, and test orchestration are unrelated to the advertised end-user function and suggest hidden maintenance or developer tooling bundled into the skill. Bundling these behaviors increases complexity and may lead to unintended local access or operational actions when the user expected only title generation.

Content

No source excerpt is available for this finding.

Self-Modification

High
Category
Rogue Agent
Confidence
85% confidence
Finding

Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.

Content

Scanner excerpt · demo_complete.py (reported line 230)May include surrounding context.

python
print("\n[LAUNCH] Next Steps:")
    print("  1. Rename skill to 'global-video-content-generator'")
    print("  2. Update SKILL.md with new features")
    print("  3. Test all content types thoroughly")
    print("  4. Set up payment system (Stripe/PayPal)")
    print("  5. Launch on ClawHub with new pricing")

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · setup.py (reported line 113)May include surrounding context.

python
*.egg

# Environment
.env
.venv
env/
venv/

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The provided manifest context names the skill 'global-video-title-generator' and describes title generation, while this file declares 'global-video-content-generator' and describes generating titles, descriptions, tags, chapters, and engagement prompts. That is a clear semantic expansion beyond title generation rather than an implementation detail.

Content

No source excerpt is available for this finding.

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 · config.json (reported line 123)May include surrounding context.

json
}
  },
  "api": {
    "base_url": "https://api.videotitlegen.com/v1",
    "rate_limits": {
      "free": {"requests_per_day": 5, "requests_per_second": 1},
      "basic": {"requests_per_month": 500, "requests_per_second": 5},

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The provided skill manifest context says this skill is a 'Global Video Title Generator' focused on title generation, but this file defines a different skill named 'global-video-content-generator' whose description is 'Affordable Video Content Generator.' It also models generation of descriptions, tags, engagement prompts, and chapters, which goes beyond title generation and indicates a semantic mismatch in intended capability.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The configuration materially diverges from the declared skill purpose by defining a broader video content product with descriptions, tags, chapters, pricing tiers, and business logic rather than a narrowly scoped title generator. This kind of scope mismatch is dangerous because it can mislead reviewers, users, or downstream systems about what the skill actually does, potentially bypassing controls, permissions review, or policy checks tied to the declared functionality.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The provided manifest context describes a 'global-video-title-generator' focused on title generation, but this file declares 'global-video-content-generator' and describes a 'Complete Video Content Generator.' It also configures generation of descriptions, tags, engagement prompts, and chapters, which is materially broader than title generation alone.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest context says the skill generates video titles, but the configuration explicitly supports 'complete' and 'description_only' modes and includes non-title outputs such as descriptions, tags, engagement prompts, and chapters. That is a semantic expansion beyond the claimed purpose of title generation.

Content

No source excerpt is available for this finding.

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 · config_v2.json (reported line 149)May include surrounding context.

json
}
  },
  "api": {
    "base_url": "https://api.videocontentgen.com/v1",
    "rate_limits": {
      "free": {"requests_per_day": 15, "requests_per_hour": 5},
      "basic": {"requests_per_month": 1500, "requests_per_hour": 10},

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code advertises the skill as supporting only English via a user-facing message. The policy allows locale constraints only when users are given a choice or when the limitation is clearly justified; neither is present here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The embedded configuration sets default_language: "en", which indicates a forced language default in natural-language/config guidance. Because no user choice or justification is provided, this conflicts with the language/locale policy.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest describes a 'Global Video Title Generator' focused on generating titles, but this file presents the skill as a 'Complete Video Content Generator' and demonstrates generation of descriptions, tags, chapters, engagement prompts, scoring, and batch content workflows. This is a semantic mismatch between the declared skill scope and the implemented/demoed behavior.

Content

No source excerpt is available for this finding.