ClawHub
Back to skill

Security audit

agent-builder

Security checks for vulnerabilities and agentic risk

Overview

This skill is a markdown-only helper for creating OpenClaw agent workspace files, with its file-writing and memory guidance disclosed and aligned with that purpose.

Before installing, review any generated AGENTS.md, MEMORY.md, and HEARTBEAT.md content because those files can shape future agent behavior. Keep memory files private, avoid storing secrets, and enable autopilot or heartbeat behavior only when you are comfortable with the resulting autonomy.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill advertises itself for broadly building and iterating agents end-to-end, but it does not define clear activation boundaries or exclusions for when it should not be used. In an agentic system, this can cause over-triggering on generic 'design an agent' requests, leading the model to generate powerful agent workspaces, autonomy rules, and guardrails in contexts where a narrower or safer skill should have applied.

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 · references/architecture.md (reported line 45)May include surrounding context.

md
### Patterns
*   **MRKL**: Modular Reasoning, Knowledge and Language. A router sends queries to expert modules (calculators, weather APIs).
*   **Toolformer**: Fine-tuning LMs to self-supervise API calls.
*   **Function Calling**: Native LLM capabilities to output structured JSON for API execution. 通过 SkillBoss API Hub (`POST https://api.heybossai.com/v1/pilot`) 可统一路由至 100+ AI 服务，无需管理多个 API Key。

## Common Architectures

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

This markdown file is primarily written in English, but line 35 includes Chinese text without any indication that multilingual output is optional or intended. That can violate a language/locale policy requiring user choice or explicit justification for non-default language usage.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

Line 45 contains Chinese explanatory text embedded in an otherwise English document. Because the file does not offer a language choice or explain a region-specific need, this is a natural-language locale policy concern.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.