Back to skill

Security audit

openclaw-collab

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed collaboration guide, but it grants broad cross-agent memory access and persistent task dispatch without enough scoping or user control.

Install only in a trusted local Hermes/OpenClaw setup. Limit use to explicit user-approved collaboration tasks, keep messages inside memory/collab where possible, avoid reading unrelated OpenClaw memory, and do not expose --json diagnostic output or delegate publishing unless you have reviewed the exact skill path and release details.

Vulnerability Patterns
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • 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
Findings (3)

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:92
Finding

Overbroad Access to OpenClaw Persistent Memory

Content
View full analysis
Remediation
View remediation

T02 · Agent Memory Poisoning

Warning
Location
SKILL.md:36
Finding

Persistent Cross-Agent Instruction Channel Without Trust Validation

Content
View full analysis
Remediation
View remediation

other

Warning
Location
SKILL.md:193
Finding

Exposure of Complete System Prompt and Agent Capability Inventory

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (7)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The file’s natural-language content, examples, and operating instructions are entirely in Chinese, and it gives prescriptive workflow guidance without indicating that another language is supported or that the user may choose their preferred locale. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
89% confidence
Finding

This is a duplicate of the persistent shared-memory behavior at line 39 and remains a real issue for the same reason: it encourages durable file-based inter-agent communication without trust boundaries. In a collaborative agent environment, such persistence can amplify prompt injection and stale-instruction risks.

Content

Scanner excerpt · SKILL.md (reported line 39)May include surrounding context.

python3.11 ~/.hermes/scripts/openclaw-bridge.py read memory/projects/

写入 OpenClaw 记忆 - 这是发送消息给OpenClaw的正确方式

python3.11 ~/.hermes/scripts/openclaw-bridge.py write memory/collab/hermes-to-openclaw.md "内容"

text

**使用场景**:

Session Persistence

Medium
Category
Rogue Agent
Confidence
89% confidence
Finding

This is a duplicate of the persistent shared-memory behavior at line 39 and remains a real issue for the same reason: it encourages durable file-based inter-agent communication without trust boundaries. In a collaborative agent environment, such persistence can amplify prompt injection and stale-instruction risks.

Content

Scanner excerpt · SKILL.md (reported line 39)May include surrounding context.

python3.11 ~/.hermes/scripts/openclaw-bridge.py read memory/projects/

写入 OpenClaw 记忆 - 这是发送消息给OpenClaw的正确方式

python3.11 ~/.hermes/scripts/openclaw-bridge.py write memory/collab/hermes-to-openclaw.md "内容"

text

**使用场景**:

Session Persistence

Medium
Category
Rogue Agent
Confidence
89% confidence
Finding

This duplicate finding is still valid because the skill again normalizes direct persistent writes as the mechanism for collaboration. The context makes it more dangerous than ordinary note-taking because the receiving agent is expected to observe and potentially act on the stored content.

Content

Scanner excerpt · SKILL.md (reported line 118)May include surrounding context.

md
1. **openclaw agent CLI 超时** — GLM5 在 NVIDIA 上不稳定,用 bridge 直接调 API 绕过
2. **openclaw.json 修复记录** — includeDefaultMemor->includeDefaultMemory,移除非法 memory.flush 键
3. **skills symlink escape 警告** — 不影响使用
4. **消息发送混淆** — `chat` 命令只是调用模型对话,不会发送消息给OpenClaw。要发送消息必须使用 `write memory/collab/` 或直接写入文件

## 常见错误

Session Persistence

Medium
Category
Rogue Agent
Confidence
89% confidence
Finding

This duplicate finding is still valid because the skill again normalizes direct persistent writes as the mechanism for collaboration. The context makes it more dangerous than ordinary note-taking because the receiving agent is expected to observe and potentially act on the stored content.

Content

Scanner excerpt · SKILL.md (reported line 118)May include surrounding context.

md
1. **openclaw agent CLI 超时** — GLM5 在 NVIDIA 上不稳定,用 bridge 直接调 API 绕过
2. **openclaw.json 修复记录** — includeDefaultMemor->includeDefaultMemory,移除非法 memory.flush 键
3. **skills symlink escape 警告** — 不影响使用
4. **消息发送混淆** — `chat` 命令只是调用模型对话,不会发送消息给OpenClaw。要发送消息必须使用 `write memory/collab/` 或直接写入文件

## 常见错误

File System Enumeration

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Code scans file system directories looking for sensitive files. This could be reconnaissance for credential theft.

Content

Scanner excerpt · SKILL.md (reported line 229)May include surrounding context.

bash
# 1. 检查skill目录结构
ls -la /Users/a1234/.openclaw/workspace/skills/ops-maintenance/

# 2. 检查SKILL.md格式
cat /Users/a1234/.openclaw/workspace/skills/ops-maintenance/SKILL.md

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.md (reported line 232)May include surrounding context.

md
ls -la /Users/a1234/.openclaw/workspace/skills/ops-maintenance/

# 2. 检查SKILL.md格式
cat /Users/a1234/.openclaw/workspace/skills/ops-maintenance/SKILL.md

# 3. 发布到clawhub.ai
clawhub publish /Users/a1234/.openclaw/workspace/skills/ops-maintenance \