澪白主控Agent
v1.0.0Provides real-time chat, intent recognition, task delegation to sub-agents, and proactive companionship through direct dialogue and agent coordination.
⭐ 0· 193·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (real-time chat, intent recognition, sub-agent dispatch, proactive companionship) aligns with the actual implementation: chat(), isTask/isEmotion intent checks, dispatch(), and active() for proactive messages. No unexpected external capabilities are requested.
Instruction Scope
SKILL.md is confined to chat/dispatch behavior. The runtime JS performs local file I/O to persist a memory file under a data/ directory (data/mem.json). This is consistent with a memory feature but is worth noting because it stores user-provided text on disk.
Install Mechanism
No install spec; the skill is instruction/code-only and does not download or execute external archives or packages during install. This is the lowest-risk install pattern.
Credentials
The skill requests no environment variables or credentials and does not access system configuration paths beyond creating/using a local ./data/mem.json file.
Persistence & Privilege
always:false and no special privileges. The skill writes a local memory file (data/mem.json) in its directory — normal for a local memory feature, but it does create persistent data on disk that may contain user messages.
Assessment
This skill appears to do what it says: chat, detect intent, dispatch simple tasks, and occasionally initiate conversation. Before installing: (1) be aware it writes conversation text to a local file at ./data/mem.json (privacy risk if the host is shared); (2) the code has small bugs (an unterminated string in memory() and an incorrect length check in keep()) that may cause runtime errors or unbounded memory growth — consider fixing those lines or asking the author for a patch; (3) run the skill in a sandboxed environment or with limited filesystem permissions if you are concerned about storing sensitive content; and (4) if you need networked sub-agent coordination, confirm whether additional code or configuration will be added (currently there are no network calls). If you want, I can point out the exact lines to fix or produce a corrected index.js copy that addresses the bugs and makes the memory file size-limit explicit.Like a lobster shell, security has layers — review code before you run it.
agentvk97b5xf9z34xg5bqewe08q2x8h82xbzpchatvk97b5xf9z34xg5bqewe08q2x8h82xbzpcompanionvk97b5xf9z34xg5bqewe08q2x8h82xbzplatestvk97b5xf9z34xg5bqewe08q2x8h82xbzp
License
MIT-0
Free to use, modify, and redistribute. No attribution required.