ClawHub
Back to skill

Security audit

SpecVibe

Security checks for vulnerabilities and agentic risk

Overview

SpecVibe is a disclosed Markdown-only development workflow skill; its broad guidance may affect many projects, but I found no hidden code, credential access, persistence, or exfiltration behavior.

Install this as a general development-process aid, not as a substitute for project-specific judgment. Be careful when applying it to regulated, safety-critical, or production deployment work, and avoid feeding private repositories into external context-packing or AI tools unless that is acceptable for your environment.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Credential Access

High
Category
Privilege Escalation
Content
### 1. Authentication (AuthN): Who are you?

- **Use Standard Protocols**: Use battle-tested standards like OAuth 2.0 or OpenID Connect (OIDC). For simple token-based auth, use JSON Web Tokens (JWT).
- **Secure JWTs**: Use a strong, randomly generated secret key (at least 256 bits) loaded from an environment variable. Set a short expiration time for access tokens (e.g., 15 minutes) and use refresh tokens for longer sessions.
- **Password Security**: NEVER store passwords in plain text. Use a strong, slow hashing algorithm like Argon2 or bcrypt.

### 2. Authorization (AuthZ): What can you do?
Confidence
70% confidence
Finding
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Session Persistence

Medium
Category
Rogue Agent
Content
SpecVibe structures development into seven distinct stages, each with a clear goal and a quality gate.

1.  **Specify**: Define user journeys and requirements in a structured `spec.md`.
2.  **Plan**: Create the technical architecture, data models, and API contracts.
3.  **Test**: Write a comprehensive, failing test suite before writing any code.
4.  **Implement**: Generate code in small, validated chunks to make the tests pass.
5.  **Review**: Perform automated and human code reviews for quality and security.
Confidence
60% confidence
Finding
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Skill Enumeration

Medium
Category
Agent Snooping
Content
```markdown
    # Project Development Framework
    @import .claude/skills/SpecVibe/SKILL.md
    ```

### For OpenClaw
Confidence
80% confidence
Finding
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Vague Triggers

Medium
Confidence
96% confidence
Finding
The skill description says to use this framework for "any new project," which is an overly broad activation phrase that can cause the skill to be invoked in contexts far beyond its validated scope. Broad auto-application increases the chance that the agent will apply generic development guidance to sensitive, regulated, or high-risk projects without requiring an explicit suitability check.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
## 3. Continuous Deployment (CD)

CD is the practice of automatically deploying your application to production after it has passed all the tests in the CI pipeline.

- **Infrastructure as Code (IaC)**: Define and manage your infrastructure using code (e.g., Terraform, Pulumi, or Docker Compose). This makes your infrastructure reproducible and version-controlled.
- **Containerization**: Package your application into a container using Docker. This ensures consistency across environments.
Confidence
80% 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.

Static analysis

No suspicious patterns detected.