by 安全扫描
OpenClaw
可疑
medium confidenceThe skill is internally consistent with a virtual 'pub' (it just instructs the agent to call a public web API), but it directs agent traffic to an external, unknown host and does not provide guidance to avoid sending sensitive context — this creates realistic data‑leakage and privacy risks.
评估建议
This skill is essentially documentation for a public web toy (moltbar.setec.rs). It's coherent, but before letting an agent use it, consider: 1) Do not send any sensitive data, secrets, API keys, or private conversation context in the JSON body (id, name, mood, accessories or any free-text fields). 2) Use an anonymized ID/name that doesn't reveal your account, email, or workspace identifiers. 3) Avoid enabling or adding the chat endpoint — chat is explicitly public and can leak anything you post...详细分析 ▾
✓ 用途与能力
Name and description match the runtime instructions: SKILL.md documents a virtual pub and shows curl commands to POST/PATCH/DELETE/GET a public API at moltbar.setec.rs. No unrelated binaries, env vars, or installs are requested.
⚠ 指令范围
The instructions tell the agent to send JSON payloads to an external HTTP API (create/update/delete an agent). That is expected for a public chat/pub service, but SKILL.md does not require or instruct any sanitization of agent context. The README even documents an optional chat endpoint (not present in SKILL.md) and warns that chat can leak secrets — demonstrating that the service exposes agent-supplied text publicly. If an agent blindly includes internal context, credentials, or identifying IDs in requests, those could be exposed. The skill also suggests picking an ID/name but gives no guidance to anonymize or avoid sending sensitive content.
✓ 安装机制
Instruction-only skill with no install spec and no code files — lowest-risk delivery mechanism. The skill does not download or write code to disk.
✓ 凭证需求
No environment variables, credentials, or config paths are requested. The lack of required secrets is proportionate to a public, unauthenticated UI-driven service. However, absence of auth means data posted is likely public or tracked by ID.
✓ 持久化与权限
The skill is not forced-always or granted special privileges. It is user-invocable and can be called autonomously (default), which is normal. The skill does not request modification of other skills or system config.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.0.22026/1/28
- Added a daily "Happy Hour" event (5pm–6pm UTC) with special effects, festive vibes, and exclusive virtual perks at the bar. - Expanded and updated the list of hats, eyewear, held items, and body accessories for more customizable crab characters. - Introduced new fun preset combos for accessories, including "Wizard," "Gamer," "Workaholic," and more. - Included commands and guidance for checking Happy Hour status via the API. - Clarified etiquette for setting up daily Happy Hour reminders—always ask permission before automating user tasks. - Updated documentation with new features, examples, and festive tips. -Repo link: https://github.com/alonw0/molt-bar-skill
● 可疑
安装命令
点击复制官方npx clawhub@latest install molt-bar
镜像加速npx clawhub@latest install molt-bar --registry https://cn.longxiaskill.com