Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Memory Auto Sync
v1.0.0Automatically records and saves all user-assistant conversations daily as Markdown files with zero dependencies and full OpenClaw compatibility.
⭐ 0· 104·1 current·1 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
SKILL.md claims the plugin only performs simple file writes ("只做纯文件写入"), but the code also writes every message into the platform memory/vector store via api.memory.store (LanceDB). Additionally the code uses a hardcoded MEMORY_DIR = "/home/tao/.openclaw/workspace/memory" instead of a user-relative or configurable path, which is unexpected for a general-purpose "zero-dependency" skill.
Instruction Scope
Runtime instructions (SKILL.md) describe real-time Markdown writes only and do not mention storing data to a vector DB. The code, however, registers listeners and dual-writes both to disk and to the platform memory API, which transmits conversation text into a vector store. SKILL.md also promises compatibility and zero complexity but the code hardcodes sender labels ("爸爸" and "张褐") and a fixed absolute path, granting it broader and non-user-configurable access than documented.
Install Mechanism
No install spec; this is an instruction-only plugin with included TypeScript extension. No remote downloads or package installs are requested. The code will be loaded by OpenClaw as an extension — expected for this type of skill.
Credentials
The skill declares no required env vars or credentials, but it unilaterally writes conversation content into the platform memory/vector store (api.memory.store) and to a hardcoded filesystem path under /home/tao. Not requesting configuration for the storage path or confirmation of the vector store behavior is disproportionate and surprising. There is also a package.json version mismatch (registry says 1.0.0, package.json 2.0.0) and no source/homepage contact information.
Persistence & Privilege
The plugin does not request always:true and registers ordinary event listeners via the plugin API. It does not modify other skills or global system settings. Autonomous invocation is allowed by default (normal).
What to consider before installing
This skill implements the stated goal (auto-saving conversations) but has multiple red flags you should resolve before installing: 1) It writes conversations into the platform memory/vector store (api.memory.store) though SKILL.md does not mention this — confirm where that data is stored, retention, and who can access it. 2) The filesystem path is hardcoded to /home/tao/... — on your machine this will likely be incorrect or could write into another user's directory; request or patch the author to use a configurable path (e.g., use the agent-provided workspace path or os.homedir()). 3) The code hardcodes sender labels which may misattribute messages; ask for clarification or modify to use actual sender metadata. 4) No author/contact/homepage and a package.json version mismatch reduce accountability. Recommended actions: review/modify the source locally (change MEMORY_DIR to a safe, configurable location or require explicit consent), test in a sandbox or non-production profile, and ask the maintainer to document the vector-store behavior and provide a configurable option for disabling api.memory.store. If you cannot obtain satisfactory answers, do not install on machines with sensitive conversations.Like a lobster shell, security has layers — review code before you run it.
latestvk97975m5g9mxffyrqrzt76wv9d8334dk
License
MIT-0
Free to use, modify, and redistribute. No attribution required.