Back to skill

Security audit

typecho-publisher

Security checks for vulnerabilities and agentic risk

Overview

This skill is mostly coherent, but it can change or permanently delete live blog content and its token handling can expose credentials if misconfigured.

Review before installing on a production blog. Use only an HTTPS Typecho URL you control, avoid passing the token with --token, store the config file with restrictive permissions, and require explicit confirmation before any update or delete action. Rotate the Typecho token if it may have been sent over HTTP, stored in logs, or shared with another agent.

Vulnerability Patterns
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
typecho-cli:20
Finding

Bearer Token and Blog Content Can Be Transmitted over Unencrypted HTTP

Content
View full analysis
环境变量 > 配置文件""" domain = None token = None # 1. 配置文件 if os.path.exists(DEFAULT_CONFIG_PATH): try: with open(DEFAULT_CONFIG_PATH, "r", encoding="utf-8") as f: cfg = json.load(f) domain = cfg.get("domain") token = cfg.get("token") except Exception: pass # 2. 环境变量 domain = os.environ.get("TYPECHO_DOMAIN", domain) token = os.environ.get("TYPECHO_TOKEN", token) # 3. 命令行参数 if getattr(args, "domain", None): domain = args.domain if getattr(args, "token", None): token = args.token if not domain or not token: print(json.dumps({ "success": False, "message": "缺少配置:请设置 domain 和 token。方式:\n" f"1. 写入配置文件 {DEFAULT_CONFIG_PATH}\n" "2. 环境变量 TYPECHO_DOMAIN / TYPECHO_TOKEN\n" "3. 命令行参数 --domain / --token" })) sys.exit(1) return domain.rstrip("/"), token def api_call(domain, token, payload): """发送 API 请求""" url = f"{domain}/index.php/action/openclaw-submit" data = json.dumps(payload).encode("utf-8") req = urllib.request.Request( url, data=data, headers={ "Content-Type": "application/json", "Authorization": f"Bearer {token}", }, method="POST", ) ``` ### Technical Analysis The CLI accepts the API domain from a configuration file, an environment variable, or a command-line argument without parsing the URL or requiring HTTPS. It then attaches the Typecho token as an HTTP bearer credential and sends the request payload to that domain. If the configured d ...[truncated 1875 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
typecho-cli:145
Finding

API Token Can Be Exposed through Process Arguments or Permissive Plaintext Configuration

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill instructs the agent to read configuration from the filesystem and environment, and to communicate with a remote blog service, but it does not declare any explicit tool scope or permission boundaries. That makes the effective authority implicit and broad, increasing the risk of secret exposure (such as TYPECHO_TOKEN), unintended file access, or network actions outside the user's expectations.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger description uses broad keywords like '发博客', '归档', '知识库', and '我的博客', which can match ordinary conversation or unrelated tasks and cause the skill to activate in contexts the user did not intend. In this skill, accidental activation is more dangerous because it can lead to real read/write/delete operations against a live blog system.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

L003 的技能说明整体以中文固定描述交互与触发条件,但未说明是否支持按用户偏好切换语言,也未给出仅限中文的业务或合规理由。根据语言/locale 政策,若技能隐含强制特定语言而无用户选择,属于自然语言策略风险。

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest explicitly advertises direct AI management of a Typecho blog, including deleting articles, but provides no warning, confirmation requirement, or safety boundary for destructive actions. In an agentic context, this increases the chance that a user request is interpreted too broadly and results in unintended content deletion or modification of production blog data.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The CLI exposes an irreversible delete operation that executes immediately once invoked, with no confirmation prompt, dry-run mode, or safety interlock. In an agent-driven skill context, this increases the chance of accidental or misinterpreted destructive actions against a live blog, especially if a user request is ambiguous or the agent is manipulated into issuing deletion commands.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

Natural-language strings including the module description, usage text, and command help are presented only in Chinese. This can violate language or locale policy when a skill forces a specific language without user opt-in or justification.

Content

No source excerpt is available for this finding.