Embedding user content in a single-quoted shell command can enable command injection
Source references: 3The create command uses the shell template `--json '<JSON parameters>'`, while title and description are derived from the user's words. The source does not require argv-based execution, stdin, or safe escaping. If an agent substitutes the JSON directly, a single quote in user content can terminate the argument and expose following text to the shell.
A malicious or accidentally crafted todo could run additional commands and potentially access or alter files, credentials, or account data available to the agent.
The documentation shows JSON inside single quotes and derives titles from user messages, but it does not include the executor or explain how the placeholder is passed. Directly interpolating user text containing a single quote into a shell string could enable command injection; an argument array or reliable escaping would prevent it. Users can ask the author to document parameter handling and restrict the agent from shell-concatenating user content.
This assessment concerns the code and conditions shown, not proof that harm has occurred.```bashwecom-cli todo create --json '<JSON 参数>'```Show 2 other places
调用本命令前,按以下规则从用户原话里提取参数。除非真的提不出,**不要**用追问让用户重新说一遍——他刚才已经把事情讲清楚了,再问一次是劣体验。绝大多数情况下能从用户消息里提炼出标题。优先采用"动宾"结构,尽量保持用户的原始表达。当标题过长,非常细节的背景细节才放进 `description`。