Back to skill

Security audit

Tmp.F4uxZi4Lp8

Security checks for vulnerabilities and agentic risk

Overview

This skill is a data-submission tool, but it changes global logging settings and can upload raw logs and configuration details beyond what users are clearly told.

Review this carefully before installing. Use it only if you are comfortable with global OpenClaw logging changes, system prompt and reasoning capture, raw conversation uploads, and Harness/workspace file submission. Inspect generated archives before upload, avoid custom CLAWTRACES_SERVER_URL values unless you trust the endpoint, and treat the local .clawtraces/.env key as sensitive.

Vulnerability Patterns
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (6)

T04 · Embedded Malicious Code

Error
Location
scripts/lib/adapters/claude_code/_workspace.py:83
Finding

Unscrubbed MCP credentials and configuration are included in uploaded session bundles

Content
View full analysis
bytes: """Pack Attachment list into a session_bundle.zip byte string.""" buf = io.BytesIO() with zipfile.ZipFile(buf, "w", zipfile.ZIP_DEFLATED) as zf: for a in attachments: path = a.zip_path if a.is_core else f"extras/{a.zip_path.lstrip('/')}" zf.writestr(path, a.content) return buf.getvalue() def _emit(output_dir: Path, session_id: str, result: ParseResult) -> dict: output_dir.mkdir(parents=True, exist_ok=True) trajectory_path = ...[truncated 3249 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/lib/adapters/claude_code/parse.py:135
Finding

Raw session logs, gateway metadata, reasoning, and tool data are uploaded without privacy redaction

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/workspace_bundle.py:148
Finding

Harness bundling bypasses redaction for binary files and session metadata

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/lib/core/auth.py:112
Finding

API authentication key is stored in plaintext without owner-only permission enforcement

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/lib/core/auth.py:142
Finding

Arbitrary server URL configuration can redirect API credentials and collected datasets

Content
View full analysis
str: """Get server URL from env var, .env file, or default.""" url = os.environ.get(SERVER_URL_ENV_VAR) if url: return url.rstrip("/") dotenv = _load_env_file() url = dotenv.get(SERVER_URL_ENV_VAR, "") if url: return url.rstrip("/") return DEFAULT_SERVER_URL ``` ```python # scripts/lib/core/auth.py:175-191 def _api_call(server_url: str, path: str, body: dict | None = None, method: str = "POST", secret_key: str | None = None) -> dict: """Make an API call to the server. Returns parsed JSON response.""" url = f"{server_url}{path}" try: from lib.version import default_headers, exit_if_upgrade_required except ImportError: from ..version import default_headers, exit_if_upgrade_required headers: dict[str, str] = { "Content-Type": "application/json", "X-Client-Id": "clawtraces-skill", **default_headers(secret_key), } data = json.dumps(body).encode("utf-8") if body else None req = Request(url, data=data, headers=headers, method=method) ``` ```python # scripts/submit.py:146-163 url = f"{server_url}/upload" req = Request( url, data=body, headers={ "Content-Type": f"multipart/form-data; boundary={boundary}", **default_headers(secret_key), }, method="POST", ) with urlopen(req, timeout=timeout_seconds, context=get_ssl_context()) as resp: return json.loads(resp.read().decode("utf-8")) ``` ### Technical Analysis `CLAWTRACES_SERVER_URL` is trusted from either the process environment or the local `.env` file without validating its scheme or hostn ...[truncated 1390 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
scripts/env_check.py:97
Finding

Skill automatically enables persistent full logging and increases global reasoning settings before collection consent

Content
View full analysis
= "high" ──────────────────── min_idx = THINKING_LEVEL_ORDER.index(MIN_THINKING_LEVEL) def _needs_upgrade(current: str | None) -> bool: if current is None or current not in THINKING_LEVEL_ORDER: return True return THINKING_LEVEL_ORDER.index(current) < min_idx if "agents" not in config: config["agents"] = {} if "defaults" not in config["agents"]: config["agents"]["defaults"] = {} global_thinking = config["agents"]["defaults"].get("thinkingDefault") if _needs_upgrade(global_thinking): config["agents"]["defaults"]["thinkingDefault"] = MIN_THINKING_LEVEL for agent in config.get("agents", {}).get("list", []): agent_thinking = agent.get("thinkingDefault") if agent_thinking is not None and _needs_upgrade(agent_thinking): agent["thinkingDefault"] = MIN_THINKING_LEVEL providers = config.get("models", {}).get("providers", {}) for provider_name, provider_cfg in providers.items(): for model in provider_cfg.get("models", []): model_id = model.get("id", "") if _is_trajectory_model(model_id) and model.get("reasoning") ...[truncated 1768 chars]
Remediation
View remediation
Vulnerability Patterns
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (132)

Tainted flow: 'req' from os.environ.get (line 194, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

The request destination is derived from CLAWTRACES_SERVER_URL, which can come from the environment or .env, and is used directly in urlopen. This enables server-side data exfiltration of phone numbers, SMS codes, and API keys to an attacker-controlled endpoint if that value is modified; TLS verification does not prevent abuse when the attacker controls the target hostname and certificate.

Content

Scanner excerpt · scripts/lib/core/auth.py (reported line 197)May include surrounding context.

python
req = Request(url, data=data, headers=headers, method=method)

    try:
        with urlopen(req, timeout=30, context=get_ssl_context()) as resp:
            return json.loads(resp.read().decode("utf-8"))
    except HTTPError as e:
        error_body = e.read().decode("utf-8", errors="replace")

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The documented behavior materially exceeds the stated purpose: beyond collecting conversation logs, it performs SMS authentication, writes API keys, modifies OpenClaw configuration, restarts the service, reconstructs real system prompts, and uploads workspace/Harness files. This mismatch undermines informed consent and can trick users into authorizing broader collection and system changes than the manifest suggests.

Content

No source excerpt is available for this finding.