AgentLove
v2.9.6AgentLove - AI 机器人一键配置（8 步完成）。用户说到"创建机器人"/"新机器人"/"配置机器人"/"备份"/"配置"/"结婚"/"进化"/"创建 agent"/"agentlove"等场景时触发。提供备份迁移、机器人配置、结婚进化三大核心功能。
⭐ 0· 168·0 current·0 all-time
byBodhi@bdodh22
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (one‑click robot configuration, backup/migration, persona presets) line up with the code and SKILL.md. The skill does not request unrelated credentials or binaries and only reads its local references/ files for persona data, which is consistent with its stated purpose. Minor metadata inconsistency: top-level listing said 'Homepage: none' while _meta.json/package.json include repository/homepage fields — likely benign bookkeeping.
Instruction Scope
SKILL.md instructs a stepwise dialog to collect configuration choices and explicitly warns users not to paste credentials in chat; code implements prompts, per‑step data collection, and in‑memory storage. The instructions reference only local reference files for persona presets (present in the package). There are no instructions to read unrelated system files or exfiltrate data to external endpoints.
Install Mechanism
No install spec and no third‑party downloads; this is an instruction‑plus-local‑code skill. No dependencies are declared in package.json. Because it is instruction-only from the platform perspective (no install hooks that fetch remote code), the install risk is low.
Credentials
The skill declares no required environment variables, no primary credential, and no config paths. SKILL.md and code consistently state credentials must be configured via the official console and the skill will not collect them. The lack of requested secrets is proportional to the described functionality.
Persistence & Privilege
always:false and state is stored only in an in‑memory Map per state-manager.js (no file writes). The skill does not request system or cross‑skill config changes. Autonomous invocation is allowed by default (platform normal) but there are no privileged operations the skill performs autonomously.
Assessment
This skill appears coherent and implements a dialog-only configuration helper that stores choices in memory and directs users to the OpenClaw console for any credentials or actual agent creation. Before installing: (1) Confirm you will not paste tokens/secrets into chat (the skill warns against this). (2) If you need stronger guarantees, ask the author how logs are handled in your deployment — the code sanitizes obvious tokens but regex‑based redaction can miss some patterns. (3) If you plan to run the included tests or inspect the package, verify network calls (there are none in the provided code) and check the referenced GitHub/homepage for the canonical source. Overall this package looks consistent with its stated purpose.Like a lobster shell, security has layers — review code before you run it.
latestvk9775w4wb4kzb3mc6kx3639gxn83d1jr
License
MIT-0
Free to use, modify, and redistribute. No attribution required.