ClawHub
Back to skill

Security audit

Thinking Protocol

Security checks for vulnerabilities and agentic risk

Overview

This skill openly changes agent-wide reasoning and response behavior, including showing internal thinking, which is risky even though no malware-like code was found.

Install only if you intentionally want this skill to affect general agent behavior. Be aware it asks the agent to reveal thinking blocks and append confidence material, which may leak sensitive context, increase tokens, and break strict JSON or other formatted outputs. Avoid copying its directives into global config unless you fully want that behavior across sessions.

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
Findings (1)

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:25
Finding

Session-Wide Reasoning and Output Instruction Hijacking

Content
View full analysis
Remediation
View remediation

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The file presents the skill examples and guidance entirely in Chinese, including the title and instructional sections, without any indication that users may choose another language. Under the policy for natural-language violations, a skill that effectively forces a specific language without opt-in should be flagged unless the locale constraint is documented and justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

The file title and all instructional content are presented in Chinese, framing the skill's usage guidance in a single language without indicating that users may choose another language. Under the policy, a skill should not impose a specific language or locale unless it offers choice or clearly documents a justified regional constraint.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
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.

Content

Scanner excerpt · SKILL-howto.md (reported line 10)May include surrounding context.

md
## 工作原理

1. OpenClaw 在会话启动时加载所有 skills
2. `skills/thinking/SKILL.md` 的内容被注入到 system prompt
3. AI 在每次推理时遵循 thinking protocol
4. 思考过程用 `thinking` 代码块包裹
5. 最终回复前进行 response preparation 检查

Ssd 3

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill explicitly instructs the agent to expose its thinking process via thinking code blocks and natural stream-of-consciousness output. Revealing intermediate reasoning can disclose internal decision processes, make prompt-injection and policy-bypass attempts easier to optimize, and increase the chance of leaking sensitive or irrelevant internal analysis to users.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

Line L248 instructs the skill to think in natural language limited to Chinese or English depending on the query. This is a natural-language locale constraint that does not offer the user an explicit language choice and is not documented as a justified region-specific limitation.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.