Security audit
Security checks for vulnerabilities and agentic risk
The skill is a local group-agent collaboration aid, but its instructions grant command/read authority and include broader file, API, and automation claims than its stated purpose clearly scopes.
Install only if you intend to let the agent run a local group-agent CLI and maintain local message archives. Keep use limited to group-management commands, review any command before execution, and avoid letting it process unrelated files or sync data to external systems unless those boundaries are clarified.
The trigger condition is so broad that it can match many ordinary planning, tracking, or collaboration requests. Overbroad activation is dangerous in a skill with exec and read permissions because it increases the chance that the agent will invoke command-capable behavior in situations where a simpler, non-privileged response would have been sufficient.
The skill demonstrates command execution, local storage, and file-backed configuration, but does not prominently warn users about the operational impact of running those actions. This is dangerous because users may treat the skill as a harmless chat-style coordination aid while it actually creates persistent state, touches the filesystem, and may execute commands with side effects.
The skill documents itself as a local-only free edition without webhook integrations, but later claims network/API capabilities and HTTPS-based communication. This kind of scope contradiction can mislead the invoking agent or user about the trust boundary, causing unintended external connectivity or overbroad tool use under a benign-looking local collaboration label.
The manifest presents a narrowly scoped group-collaboration tool, but the body expands into generic file processing, API integration, information retrieval, and command automation. This is dangerous because policy engines, users, or agents may grant trust based on the narrow manifest while the prose authorizes much broader operations, increasing the chance of unintended execution or data exposure.
The skill claims broad file-processing support even though its declared purpose is lightweight group messaging. Unjustified file-handling claims increase risk because they normalize access to local data and may encourage an agent to read or manipulate unrelated files outside the collaboration task scope.
The documentation claims external API integration despite the manifest framing this as a local group-collaboration utility. This mismatch is dangerous because it expands the effective attack surface to outbound data transfer and remote interactions without a corresponding declaration of trust, authentication, or consent boundaries.
描述中将“支持中文交互”作为显著行为特征，但没有同时说明是否支持其他语言、是否允许用户自行选择输出语言。根据自然语言政策要求，若技能对语言/locale 存在偏向或约束，应提供用户选择或明确的合理限定。
