Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
LLM Knowledge Base
v1.0.0Create and manage LLM-powered personal knowledge bases using Karpathy's wiki method. Use when users want to build a knowledge base, ingest sources, query the...
⭐ 1· 14·0 current·0 all-time
byGavin C.@gavinchengcool
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description promise a local LLM-backed knowledge base and the SKILL.md only asks the agent to create a directory structure, write Obsidian config and markdown files, ingest sources, and update wiki pages. There are no unrelated environment variables, binaries, or install steps requested.
Instruction Scope
Instructions stay largely within the KB-building scope (mkdir, write templates, copy files, fetch URLs with curl, read/update wiki pages). The only potentially sensitive operation is fetching external URLs (web search + curl) and ingesting user-provided local files; this is expected for an ingesting KB skill but you should be aware it involves network I/O and reading user-supplied files.
Install Mechanism
This is instruction-only with no install spec or bundled code. No archives or external installers are referenced, so there's no install-time code being dropped by the skill itself.
Credentials
The skill declares no required environment variables or credentials. It does instruct reading/writing files in the created vault and optionally fetching external URLs; those capabilities are proportionate to a knowledge-base ingestion and management task, but they mean the agent will access any files you point it to.
Persistence & Privilege
always is false and the skill does not request persistent platform privileges. It only describes creating and updating files within the vault it creates; it does not modify other skills or system-wide agent settings.
Assessment
This skill is internally consistent with its purpose, but be cautious about what you let it ingest. Practical precautions: 1) Run the skill where you want the vault created (it writes files relative to the current directory). 2) Do not drop secrets or sensitive system files into raw/inbox — anything you point the skill at may be read and stored. 3) Fetching URLs (web search + curl) means network access; avoid using it in environments where outward requests could leak internal URLs or trigger SSRF-like behavior. 4) Review the generated AGENTS.md and any new files before trusting saved outputs. If you want extra safety, run the initial creation in an isolated folder or sandbox and inspect created files and fetched source content before allowing further automated ingestion or sharing.Like a lobster shell, security has layers — review code before you run it.
latestvk97447fbatjf0qb97j98t4y73s84bjj6
License
MIT-0
Free to use, modify, and redistribute. No attribution required.