Back to skill

Security audit

bbc-skill-tool

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent WhatsApp bot-building guide with cloud bot mutation and deployment powers, but its sensitive behavior is mostly disclosed and purpose-aligned.

Before installing, be comfortable giving the agent authority to create, update, validate, deploy, reboot, and delete BuilderBot Cloud bot resources. Review any CRM, dashboard, webhook, website-scraping, or media URL configuration before use, and add your own privacy notices, consent, retention, and access-control rules when collecting customer details or forwarding conversation context.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (12)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 374)May include surrounding context.

md
## ANTI-PATTERNS (Never Do These)

1. ❌ Creating a flow without immediately adding answers
2. ❌ Reporting success without verification (`list_*`)
3. ❌ Deleting without showing impact and getting confirmation
4. ❌ Deploying without running `validate_bot` first
5. ❌ Mixing `add_chatpdf` and `add_text` in the same flow

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 · SKILL.md (reported line 158)May include surrounding context.

message: "" plugins: http: url: "https://api.example.com/status?order={capture}" method: "GET" rules: [] ← REQUIRED, even if empty

text

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 · advanced-patterns.md (reported line 12)May include surrounding context.

message: "" plugins: http: url: "https://api.example.com/status?order={capture}" method: "GET" rules: [] ← REQUIRED, even if empty

text

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 · advanced-patterns.md (reported line 24)May include surrounding context.

message: "" plugins: http: url: "https://api.example.com/status?order={capture}" method: "GET" rules: [] ← REQUIRED, even if empty

text

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The example explicitly collects user identifiers such as name and email, then posts the captured data to an external CRM without any guidance to obtain consent, disclose third-party sharing, minimize fields, or protect sensitive data. In a bot-building skill, this pattern is likely to be copied directly into production, creating a real privacy and compliance risk rather than a purely theoretical concern.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The handoff pattern advises posting conversation context to an agent dashboard but does not warn that message history may contain personal, financial, or health-related information. Because this is presented as an advanced support pattern for WhatsApp business bots, operators may forward entire transcripts to third parties without user awareness or filtering.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The instruction "Responde siempre en español" imposes a fixed language policy on users. This is a natural-language policy issue because the template does not provide any opt-in, fallback, or choice for users who may prefer another language.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The line "Responde en español" mandates a specific language for all interactions. Because the template does not mention user choice or a justified regional restriction, it violates the language/locale policy criteria.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This instruction requires the assistant to answer in Spanish regardless of user preference. The file does not document a justified locale limitation or offer an opt-in choice, so it fits the policy-violation category.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The template directs the assistant to always respond in Spanish, but does not give users a language option. That creates a language policy concern under the natural-language policy rules.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This rule mandates Spanish responses for the assistant without any opt-in or user choice. Since no justified regional constraint is stated, this is a language/locale policy violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

The markdown provides a direct example for deleting a knowledge base file but does not warn that the action is destructive or potentially irreversible. For markdown files, operations affecting user data or system integrity should be accompanied by warnings.

Content

No source excerpt is available for this finding.