Back to skill

Security audit

久吾智能体

Security checks for vulnerabilities and agentic risk

Overview

The skill is purpose-aligned for Jiuwu document review, but it sends tokens and potentially confidential text or files over a hardcoded plaintext HTTP endpoint, so users should review it carefully before installing.

Install only if you expect to use this inside the intended Jiuwu/internal network and are comfortable sending selected text and files to that service. Before using it with confidential contracts or business documents, require an HTTPS endpoint, avoid sending document text in URL query strings, confirm the hardcoded 192.168.1.213:5000 destination is correct, and pin reviewed dependency versions.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/call_agent.py:57
Finding

Bearer Token and Confidential Documents Transmitted over Plaintext HTTP

Content
View full analysis
dict: """调用文本分析API""" encoded_content = urllib.parse.quote(content) url = (f"{BASE_URL}/api/AiReview/AgentAnswer" f"?name={urllib.parse.quote(name)}" f"&docno={urllib.parse.quote(docno)}" f"&content={encoded_content}") request = urllib.request.Request(url, method='GET') request.add_header('accept', 'text/plain') request.add_header('Authorization', f'Bearer {AUTH_TOKEN}') with urllib.request.urlopen(request, timeout=timeout) as response: return json.loads(response.read().decode('utf-8')) def call_file_api(name: str, docno: str, files: list, code: str = "", timeout: int = DEFAULT_TIMEOUT) -> dict: """调用文件分析API""" import requests url = f"{BASE_URL}/api/AiReview/AgentAnswerByFiles" files_data = [] for file_path in files: with open(file_path, 'rb') as f: file_content = f.read() files_data.append(('files', (os.path.basename(file_path), file_content, 'application/octet-stream'))) data = { 'code': code, 'name': name, 'docno': docno } headers = { 'accept': 'text/plain', 'Authorization': f'Bearer {AUTH_TOKEN}' } response = requests.post(url, data=data, files=files_data, headers=headers, timeout=timeout) return response.json() ``` ### Technical Analysis Both API operations use a hardcoded `http://` endpoint rather than HTTPS. The requests carry a bearer token and potentially confidential contract text, document identifiers, and complete uploaded files. Plain HTTP p ...[truncated 2632 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
scripts/requirements.txt:1
Finding

Dependency Version Is Not Reproducibly Pinned

Content
View full analysis
=2.25.0 ``` ### Technical Analysis The `requests` dependency has only a minimum version constraint. Every future release satisfying `>=2.25.0` may be installed, including versions that were not reviewed with this Skill. This makes installations non-reproducible and can introduce unexpected behavior, compatibility regressions, or newly compromised dependency artifacts. The package name is legitimate, and the project does not configure a suspicious package index. Therefore, this finding does not establish that a malicious package is currently present. The risk arises from allowing future, unaudited versions to enter the environment automatically. ### Attack Path 1. A user or deployment system installs the requirements at a later date. 2. The package resolver selects any available `requests` version satisfying the lower-bound constraint. 3. A compromised, maliciously replaced, or behaviorally incompatible eligible release is installed. 4. The installed package executes within the Skill's Python environment and participates in authenticated file-upload requests. 5. A compromised dependency could access request headers and uploaded data available to the process or alter network behavior within the process's existing privileges. ### Impact Assessment The dependency runs with the same local privileges as the Skill process. If a selected package release were compromised, it could potentially access process-visible environment variables, including `JIUWU_CORE_TOKEN`, read files available to the invoking user, inspect uploaded document contents, or alter outgoing requests. The exact impact depends on the privileges of the account running the Skill and the availability of a compromised eligible release. No current dependency compromise was identified during th ...[truncated 13 chars]
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (15)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill description and usage instructions say it analyzes text and files through the Jiuwu agent API, but they do not prominently warn users that submitted text and uploaded files will be transmitted to an external third-party service. This undermines informed consent and can expose confidential contracts, requirements, or internal documents to external processing without adequate user awareness.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/call_agent.py (reported line 9)May include surrounding context.

python
Token 读取优先级:
1. OpenClaw 环境变量 JIUWU_CORE_TOKEN (env.vars)
2. workspace/.env 文件
3. OpenClaw根目录的 .env 文件
"""

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/call_agent.py (reported line 10)May include surrounding context.

python
Token 读取优先级:
1. OpenClaw 环境变量 JIUWU_CORE_TOKEN (env.vars)
2. workspace/.env 文件
3. OpenClaw根目录的 .env 文件
"""

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/call_agent.py (reported line 44)May include surrounding context.

python
Token 读取优先级:
1. OpenClaw 环境变量 JIUWU_CORE_TOKEN (env.vars)
2. workspace/.env 文件
3. OpenClaw根目录的 .env 文件
"""

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/call_agent.py (reported line 48)May include surrounding context.

python
Token 读取优先级:
1. OpenClaw 环境变量 JIUWU_CORE_TOKEN (env.vars)
2. workspace/.env 文件
3. OpenClaw根目录的 .env 文件
"""

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/call_agent.py (reported line 32)May include surrounding context.

python
if token:
        return token

    workspace_env = Path.home() / '.openclaw' / 'workspace' / '.env'
    if workspace_env.exists():
        token = parse_env_file(workspace_env)
        if token:

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/call_agent.py (reported line 38)May include surrounding context.

python
if token:
        return token

    workspace_env = Path.home() / '.openclaw' / 'workspace' / '.env'
    if workspace_env.exists():
        token = parse_env_file(workspace_env)
        if token:

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill explicitly instructs use of a Python script, reads a token from environment variables, accepts local file paths, and sends content to an external API, yet it declares no tool scope or permissions boundary. This creates an authorization and transparency gap: an agent may invoke network, file-read, and env-backed behavior without an explicit manifest-level constraint or review signal.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The trigger description is broad enough to match many ordinary requests involving contracts, clauses, document review, or general text/file analysis, increasing the chance the skill is auto-selected in situations the user did not intend. Because the skill transmits user text or files to an external service, overbroad triggering can cause unintended data disclosure or use of the wrong external processor.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

User-facing strings including the module description, argument help, status messages, and error text are all hard-coded in Chinese. There is no option for users to select another language and no documented justification that the tool is intentionally restricted to a Chinese-language environment.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The script automatically loads a bearer token from environment variables and local .env files, but the skill description only describes document analysis behavior and does not disclose credential access. This is dangerous because users may invoke the skill without realizing it reads local secrets and uses them to authenticate outbound requests, expanding the trust boundary beyond what was advertised.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The code sends arbitrary text and uploaded file contents to a remote API endpoint, including potentially sensitive contracts or internal documents, without any user-facing warning or confirmation. In this skill context, the data being analyzed is likely confidential, so silent transmission to an external service materially increases privacy and data leakage risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The script silently reads authentication material from environment variables and .env files without disclosing this behavior to users. While the code does not exfiltrate the token directly except for intended API authentication, undisclosed secret handling is still risky because it can surprise users and normalize broader credential access than the skill description suggests.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
96% confidence
Finding

The dependency is specified as requests>=2.25.0, which allows installation of many different future or environment-dependent versions rather than a reviewed, reproducible release. This creates supply-chain and patch-verification risk because builds may resolve to different versions, and security review cannot confirm whether a vulnerable or safe release will actually be installed.

Content

Scanner excerpt · scripts/requirements.txt (reported line 1)May include surrounding context.

text
requests>=2.25.0

Unverifiable Dependency: requests has 16 known advisory(ies) (CVE-2014-1830 (Exposure of Sensitive Information to an Unauthorized Actor in Requests); CVE-2024-47081 (Requests vulnerable to .netrc credentials leak via malicious URLs); CVE-2024-35195 (Requests `Session` object does not verify requests after making first request wi) +13 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
93% confidence
Finding

Because requests is not pinned, it is impossible to determine from this manifest whether deployment will use a version affected by one of the known advisories for that package. In a skill that sends text or files to an external API, dependency behavior matters for transport security and credential handling, so unresolved package versioning slightly increases operational risk.

Content

No source excerpt is available for this finding.