Back to skill

Security audit

Openclaw Multi Agents

Security checks for vulnerabilities and agentic risk

Overview

The skill is related to OpenClaw team setup, but it makes broad persistent changes to agent behavior and local workspaces with unsafe installation and data-handling patterns.

Install only after reviewing the generated SOUL.md, AGENTS.md, and openclaw.json diffs, and only if you want your Main Agent to persistently delegate work through a Manager. Avoid live prompts.chat enrichment, pin or manually review added skills, back up ~/.openclaw first, and tighten workspace permissions before using this on any shared machine.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (6)

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:438
Finding

Persistent Main-Agent Role and Workflow Hijacking

Content
View full analysis
Remediation
View remediation

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:382
Finding

Untrusted Remote Prompt Content Persisted as Agent Instructions

Content
View full analysis
&perPage=3" ``` | Responsibility | Search Term | |------|--------| | Feature Development | `software engineer` | | Code Review | `code reviewer` | | Strategic Planning | `product manager` | | Data Analysis | `data analyst` | | Marketing Copy | `copywriter` | Extract **working principles and output formats** from the `content` field of the returned results and integrate them into the corresponding sections of the template below: ``` ### Technical Analysis The Skill fetches prompt content from an external service and directs the agent to integrate the returned `content` field into persistent Worker `SOUL.md` files. The remote response is neither pinned nor cryptographically verified, and no sanitization or instruction-boundary enforcement is defined. Prompt content is executable in the semantic sense once placed into an agent's core instruction file. A malicious or compromised API response could introduce directives to use tools, disclose data, bypass safety checks, communicate with unauthorized agents, or modify additional files. The repository can therefore behave differently after audit without any local source change. The effective Worker instruction payload depends on mutable third-party content at installation time. ### Attack Path 1. An attacker compromises the remote API, a returned prompt entry, or the service's content pipeline. 2. The API returns content containing malicious behavioral or tool-use instructions. 3. The Skill treats the response a ...[truncated 923 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Error
Location
scripts/install_skills.sh:24
Finding

Unpinned Third-Party Skill Installation

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/create_agent.sh:7
Finding

Path Traversal Through Unvalidated Agent Identifiers

Content
View full analysis
" echo "Example: $0 munger" exit 1 fi echo "🚀 Creating sub-agent: $AGENT_NAME" mkdir -p "$AGENT_DIR"/{memory,skills} touch "$AGENT_DIR"/{SOUL.md,AGENTS.md,IDENTITY.md,WORKSPACE.md} mkdir -p "$AGENT_DIR/memory" echo "# Memory Index" > "$AGENT_DIR/memory/index.md" mkdir -p "$AGENT_DIR/skills" cat > "$AGENT_DIR/WORKSPACE.md" << EOF # WORKSPACE.md 这是 ${AGENT_NAME} 的独立工作空间. EOF chmod -R 755 "$AGENT_DIR" ``` Equivalent unsafe path construction appears elsewhere: ```bash AGENT_DIR=~/.openclaw/workspace-$AGENT_NAME MANAGER_DIR=~/.openclaw/workspace-$MANAGER_ID local MANAGER_DIR="$HOME/.openclaw/workspace-$MANAGER_ID" local WORKER_DIR="$HOME/.openclaw/workspace-$worker_id" ``` ### Technical Analysis Agent, Manager, and Worker identifiers are incorporated into filesystem paths without validation. The scripts check only whether values are present; they do not restrict slashes, `..` components, absolute-path syntax, control characters, or other path-manipulation input. Quoting prevents shell word splitting but does not prevent path traversal. An identifier containing path separators and parent-directory components can cause the resolved path to leave the intended `$HOME/.openclaw/workspace-*` namespace. After constructing the path, the scripts create directories, overwrite known filenames, and recursively apply permissions. The combination increases impact beyond simple directory creation. ### Attack Path 1. An attacker persuades a user or automation layer to run a setup script with a crafted agent identifier containing traversal ...[truncated 1117 chars]
Remediation
View remediation
&2 exit 1 fi ``` - Apply the same validation to Manager IDs, Main Agent IDs, Worker IDs, and team-generated identifiers. - Build paths from a fixed base directory and canonicalize both the base and destination. - Verify that the canonical destination is an immediate child of the intended base before creating or writing anything. - Reject all identifiers containing `/`, `\`, `..`, shell control characters, or leading hyphens. - Refuse to operate if the destination is a symbolic link. - Avoid recursive permission changes. - Add regression tests for traversal strings, absolute paths, whitespace, Unicode separators, and symlink targets. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/run_planning_interview.sh:24
Finding

Predictable Shared Temporary File for Interview Data

Content
View full analysis
"$PROGRESS_FILE" } ``` ```bash load_progress() { if [ -f "$PROGRESS_FILE" ]; then print_warning "发现未完成的访谈记录" echo -e " 时间: $(grep -o '"timestamp":"[^"]*"' "$PROGRESS_FILE" | cut -d'"' -f4)" echo "" read -p "是否继续上次进度? [Y/n]: " resume_choice if [[ "$resume_choice" =~ ^[Nn]$ ]]; then rm -f "$PROGRESS_FILE" print_success "已清除旧记录,开始新访谈" return 1 fi CURRENT_QUESTION=$(grep -o '"current_question":[0-9]*' "$PROGRESS_FILE" | cut -d: -f2) local answers_raw=$(grep -o '"answers":{.*}' "$PROGRESS_FILE" | sed 's/"answers"://') fi } ``` ### Technical Analysis The script stores interview answers in a fixed, globally predictable path under `/tmp`. It does not use `mktemp`, exclusive creation, ownership validation, restrictive `umask`, or symbolic-link checks. On multi-user systems, another local account may pre-create the path, monitor it, replace it, or make it a symbolic link. The interview asks about user workflows, pain points, task types, and team plans, so the file can contain sensitive business or personal context. The manually generated JSON als ...[truncated 1187 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/create_agent.sh:52
Finding

Agent Workspaces and Memory Are Made World-Readable

Content
View full analysis
/dev/null || true ``` ```bash chmod -R 755 "$MANAGER_DIR" ``` ```bash chmod -R 755 "$WORKER_DIR" ``` ### Technical Analysis The setup scripts recursively assign mode `755` to entire agent workspaces. This makes directories traversable and files readable by every local account. It also marks ordinary data files as executable because the execute bit is applied recursively. The affected workspaces contain persistent prompts, identity definitions, memory, task context, and potentially user-specific operational information. These files do not require global read access to perform the stated task. Recursive permission modification can also weaken previously restrictive files if the setup script is run against an existing workspace. ### Attack Path 1. A user creates or updates an agent with one of the setup scripts. 2. The script writes identity, prompt, workspace, and memory files. 3. The script recursively changes the entire workspace to mode `755`. 4. Another local account traverses the workspace and reads the generated files. 5. If the workspace already contained sensitive files, those files also become readable. ### Impact Assessment Other local users can potentially read: - `SOUL.md` and `AGENTS.md` behavioral instructions. - `IDENTITY.md` and `WORKSPACE.md`. - Agent memory and task history. - Installed Skill content. - User context copied into agent-generated files. This issue does not grant write access to other users under ordinary ownership rules, but it causes unnecessary local information disclosure and broadens attack reconnaissance opportunities. ]]>
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Rogue AgentSelf-Modification, Session Persistence
Findings (71)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

This duplicate scanner hit points to the same destructive workspace deletion command. It remains a true issue because the documentation normalizes direct recursive removal of persistent agent data without any protective framing.

Content

Scanner excerpt · INSTALL.md (reported line 659)May include surrounding context.

md
openclaw gateway restart

# 3. 可选:删除 workspace(会同时删除 SOUL.md 和记忆)
rm -rf ~/.openclaw/workspace-manager
rm -rf ~/.openclaw/workspace-munger

# 4. 可选:删除 session 历史

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

This duplicate scanner hit points to the same destructive workspace deletion command. It remains a true issue because the documentation normalizes direct recursive removal of persistent agent data without any protective framing.

Content

Scanner excerpt · INSTALL.md (reported line 659)May include surrounding context.

md
openclaw gateway restart

# 3. 可选:删除 workspace(会同时删除 SOUL.md 和记忆)
rm -rf ~/.openclaw/workspace-manager
rm -rf ~/.openclaw/workspace-munger

# 4. 可选:删除 session 历史

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

This duplicate finding covers the same worker-workspace deletion command and carries the same accidental-data-loss risk. The skill context makes it more sensitive because these directories are explicitly used to store durable memory and agent configuration.

Content

Scanner excerpt · INSTALL.md (reported line 660)May include surrounding context.

md
# 3. 可选:删除 workspace(会同时删除 SOUL.md 和记忆)
rm -rf ~/.openclaw/workspace-manager
rm -rf ~/.openclaw/workspace-munger

# 4. 可选:删除 session 历史
rm -rf ~/.openclaw/agents/manager

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

This duplicate finding covers the same worker-workspace deletion command and carries the same accidental-data-loss risk. The skill context makes it more sensitive because these directories are explicitly used to store durable memory and agent configuration.

Content

Scanner excerpt · INSTALL.md (reported line 660)May include surrounding context.

md
# 3. 可选:删除 workspace(会同时删除 SOUL.md 和记忆)
rm -rf ~/.openclaw/workspace-manager
rm -rf ~/.openclaw/workspace-munger

# 4. 可选:删除 session 历史
rm -rf ~/.openclaw/agents/manager

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

This duplicate hit refers to the same session-directory deletion command. Even though it is not an exploit primitive in itself, it is operationally dangerous because it permanently erases hidden state from the user's home directory.

Content

Scanner excerpt · INSTALL.md (reported line 663)May include surrounding context.

rm -rf ~/.openclaw/workspace-munger

4. 可选:删除 session 历史

rm -rf ~/.openclaw/agents/manager rm -rf ~/.openclaw/agents/munger

text

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

This duplicate hit refers to the same session-directory deletion command. Even though it is not an exploit primitive in itself, it is operationally dangerous because it permanently erases hidden state from the user's home directory.

Content

Scanner excerpt · INSTALL.md (reported line 663)May include surrounding context.

rm -rf ~/.openclaw/workspace-munger

4. 可选:删除 session 历史

rm -rf ~/.openclaw/agents/manager rm -rf ~/.openclaw/agents/munger

text

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

This duplicate hit refers to the same worker session-history deletion command. The risk is unchanged: irreversible loss of state and history from copy-pasted destructive shell commands.

Content

Scanner excerpt · INSTALL.md (reported line 664)May include surrounding context.

4. 可选:删除 session 历史

rm -rf ~/.openclaw/agents/manager rm -rf ~/.openclaw/agents/munger

text

---

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

This duplicate hit refers to the same worker session-history deletion command. The risk is unchanged: irreversible loss of state and history from copy-pasted destructive shell commands.

Content

Scanner excerpt · INSTALL.md (reported line 664)May include surrounding context.

4. 可选:删除 session 历史

rm -rf ~/.openclaw/agents/manager rm -rf ~/.openclaw/agents/munger

text

---

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The declared description promises a sophisticated multi-agent system inside OpenClaw: planning/delegation by a Manager, specialized Workers, mandatory QA on outputs, and numerous persona choices. The supplied code does not implement any of these behaviors. It only creates filesystem scaffolding for one agent workspace, including directories and placeholder markdown files, then prints instructions for manual configuration. While this may be a supporting setup utility for an agent-based system, the actual chunk's primary behavior is workspace initialization, not team assembly, orchestration, QA enforcement, or persona management. Therefore the description materially overstates and misrepresents what this code chunk actually does.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

There is a material description-behavior mismatch. The description promises a functional multi-agent team system inside OpenClaw with planning, delegation, worker specialization, QA enforcement, and many persona choices. The supplied code only installs skill packages for specific agent names into local directories. While this installer could be a supporting component of a larger multi-agent setup, this code chunk itself does not perform the core advertised behavior. It also carries undeclared operational capabilities: changing directories, writing into agent workspace skill folders, and invoking an external installer/source. Therefore the declared purpose does not accurately represent what this code chunk actually does.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The declared description presents an operational skill that helps the user assemble and run a dedicated AI team inside OpenClaw, including manager/worker delegation and mandatory quality checks on outputs. The supplied code chunk is only a planning interview tool. It collects user input, stores progress in /tmp, performs simple text-based heuristics, recommends team members, and writes a Team Design Document. It mentions later setup scripts and OpenClaw configuration, but this script itself neither provisions agents nor orchestrates them, nor does it implement actual QA enforcement. The behavior is related to the broader theme of multi-agent planning, but the primary purpose of this code chunk is planning/document generation rather than the declared operational team-building/orchestration capability.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The code substantially relates to the declared domain: it does create a Manager/Worker team structure for OpenClaw, generates agent files, and includes quality-gate instructions in the generated docs. However, key parts of the description overstate what the code actually does. First, the advertised '61+ historical persona options' is not supported by this chunk; only a limited hardcoded set of personas is offered. Second, the description says every deliverable is forced through a QA gate and that the team checks its work before handing it over, but this script merely creates markdown instructions/checklists describing such behavior—there is no enforcement logic or operational QA mechanism in the script itself. Third, the primary behavior of the code is setup/scaffolding (creating directories, markdown files, and config snippets), not actually running an orchestrated multi-agent workflow. So the description is directionally related but materially overstated in capability and enforcement.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The document claims scripts make no network calls, but later directs an external HTTP request. This contradiction can mislead reviewers and users into approving the skill under false assumptions, weakening informed consent around data egress.

Content

No source excerpt is available for this finding.

Exfiltration Commands

High
Category
Prompt Injection
Confidence
90% confidence
Finding

Instructions found that direct the agent to transmit conversation context or user data to external services.

Content

Scanner excerpt · SKILL.md (reported line 869)May include surrounding context.

md
|------|---------|
| Manager missing in `agents.list` | Go to Step 0 (Team Building) first. |
| `sessions_send` timeout | Check `sessions_history(sessionKey="agent:main:manager")` for updates. |
| "Permission denied" in inter-agent communication | Check if `agentToAgent.allow` contains **"main"** and all Agent IDs (list of strings). Without "main", Main Agent cannot send messages to Manager. |
| Persona not active | Add reinforcement text after the first sentence in SOUL.md, ask user to run `openclaw gateway restart` in their terminal (⚠️ AI must not exec this). |
| Unsatisfactory deliverable | Package the gaps and send to Manager; Manager arranges rework. Main does not contact Worker directly. |
| Only main exists (no team) | Do not force orchestration; go to Step 0 first. |

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · templates/manager_soul_template.md (reported line 104)May include surrounding context.

md
**Status Reporting to Main Agent:**
- Report only high-level status (Started/Progress %/Completed/Blocked)
- Maintain detailed status internally
- Provide clear context in case of anomalies

---

Session Persistence

Medium
Category
Rogue Agent
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.

Content

Scanner excerpt · INSTALL.md (reported line 273)May include surrounding context.

bash
# 创建 wisdom 目录并复制样例文件
mkdir -p ~/.openclaw/workspace/memory/wisdom
SKILL_DIR="$(openclaw skills info openclaw-multi-agents --json 2>/dev/null | grep path | head -1 | sed 's/.*: "\(.*\)".*/\1/' || find ~/.openclaw/workspace/skills -name "SKILL.md" | grep multi-agents | xargs dirname)"
if [ -n "$SKILL_DIR" ] && [ -d "$SKILL_DIR/examples/wisdom" ]; then
  cp "$SKILL_DIR/examples/wisdom/"*.md ~/.openclaw/workspace/memory/wisdom/

Session Persistence

Medium
Category
Rogue Agent
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.

Content

Scanner excerpt · INSTALL.md (reported line 430)May include surrounding context.

bash
# 创建 Manager workspace
mkdir -p ~/.openclaw/workspace-manager

# 创建 Manager SOUL.md(参考 docs/team_templates.md)
# 创建 Manager AGENTS.md

Session Persistence

Medium
Category
Rogue Agent
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.

Content

Scanner excerpt · INSTALL.md (reported line 430)May include surrounding context.

bash
# 创建 Manager workspace
mkdir -p ~/.openclaw/workspace-manager

# 创建 Manager SOUL.md(参考 docs/team_templates.md)
# 创建 Manager AGENTS.md

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The uninstall section instructs users to recursively delete workspaces and session-history directories, but it does not present an explicit data-loss warning, backup guidance, or confirmation step. In a skill that promotes persistent multi-agent memory and stored sessions, these directories likely contain valuable state, prompts, and history, so users could irreversibly destroy important data by following the commands casually.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The documented auto-activation phrases are broad, natural-language requests like asking to build a team or help with programming, which can overlap with ordinary conversation. In an agent platform, this can cause unintended skill invocation, leading to unexpected delegation, background task creation, or access to additional context and memory without the user explicitly intending to use this skill.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README states that the skill reads existing context including USER.md, memory, and session history, but it does not present a prominent privacy notice, scope limitation, or consent mechanism near that behavior. This creates a data exposure risk because users may trigger the skill without realizing it will ingest prior conversations or stored memory, especially when combined with the broad activation language described earlier.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The README states the skill will 'Read your existing context (USER.md, memory, session history)' before proceeding, but it does not provide a visible privacy warning, scope limitation, or consent model. Because this skill is specifically designed for orchestration and background work inside OpenClaw, silent access to historical context increases the chance of over-collection and user surprise, especially if sensitive material exists in session history or memory stores.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The README advertises broad natural-language triggers such as 'I want to build a team' and 'Help me handle multiple things at once' and states the skill 'activates automatically.' In a chat-driven agent platform, this creates ambiguous invocation boundaries and raises the risk of unintended activation during ordinary conversation, which could cause the skill to read context or begin orchestration workflows without sufficiently explicit user intent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README describes persistent storage of 'Wisdom' derived from tasks in shared workspace memory without a user-facing warning about retention, reviewability, or sensitive-data handling. In a multi-agent system, accumulated notes can capture confidential project details, mistakes, conventions, or operational insights and then be automatically reinjected into future tasks, amplifying the privacy and data leakage risk across sessions and agents.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.